GluN3A expression restricts spine maturation via inhibition of GIT1/Rac1 signaling

Abstract

NMDA-type glutamate receptors (NMDARs) guide the activity-dependent remodeling of excitatory synapses and associated dendritic spines during critical periods of postnatal brain development. Whereas mature NMDARs composed of GluN1 and GluN2 subunits mediate synapse plasticity and promote spine growth and stabilization, juvenile NMDARs containing GluN3A subunits are thought to inhibit these processes via yet unknown mechanisms. Here, we report that GluN3A binds G protein-coupled receptor kinase-interacting protein (GIT1), a postsynaptic scaffold that assembles actin regulatory complexes, including the Rac1 guanine nucleotide exchange factor βPIX, to promote Rac1 activation in spines. Binding to GluN3A limits the synaptic localization of GIT1 and its ability to complex βPIX, leading to decreased Rac1 activation and reduced spine density and size in primary cultured neurons. Conversely, knocking out GluN3A favors the formation of GIT1/βPIX complexes and increases the activation of Rac1 and its main effector p21-activated kinase. We further show that binding of GluN3A to GIT1 is regulated by synaptic activity, a response that might restrict the negative regulatory effects of GluN3A on actin signaling to inactive synapses. Our results identify inhibition of Rac1/p21-activated kinase actin signaling pathways as an activity-dependent mechanism mediating the inhibitory effects of GluN3A on spine morphogenesis.

Keywords: actin cytoskeleton; structural plasticity; synaptic refinement.

Fig. 1.

GluN3A binds the synaptic scaffold GIT1. (A) Two independent clones encoding GIT1 were isolated by yeast two-hybrid screening using the intracellular carboxyl-terminal domain of GluN3A (GluN3A Ct, amino acids 952–1,115) as bait. Interactions were tested by induction of the reporter genes LacZ (β-Gal) and His3. (B) Amino acids 1,082–1,115 in the distal GluN3A tail were sufficient for GIT1 binding. Amino acid numbers refer to positions C-terminal to the last transmembrane domain. (C) Whole extracts of P8 mouse forebrain were incubated with control GST beads or beads bound to GST fused to the entire GluN3A Ct or smaller fragments of the Ct as indicated. Precipitated proteins were detected by immunoblot (IB) using an antibody to GIT1, and Coomassie staining was used to visualize GST-fusion proteins in the same gel. (D) Diagram of GIT1 and truncations used. ArfGAP, ADP ribosylation factor-GAP domain; PBD, paxillin-binding domain; SHD, Spa2 homology domain; SLD, synaptic localization factor. (E) Extracts of HEK 293 cells transfected with GluN1 plus the indicated constructs were incubated with anti-GFP antibody, and immunoprecipitated proteins were analyzed by IB with the indicated antibodies. GluN3A full-length, or lacking amino acids 1,082–1,115 responsible for GIT1 binding (GluN3AΔGIT1), were tagged with GFP; entire GIT1 or GIT1 lacking the synaptic localization domain (GIT1ΔSLD) was Flag-tagged. Input is 5% of the lysate used for the immunoprecipitate (IP). (F) Solubilized mouse forebrain membrane extracts were incubated with GluN3A antibody or control IgG. Rb, rabbit. (G) Lysates from WT and GluN3A^−/− P8 mouse forebrain were incubated with GluN1 antibody or control IgG, and immunoprecipitated proteins were analyzed by IB. Mo, mouse. In F and G, input is 10% of the lysate used for immunoprecipitation.

Fig. 2.

Inhibition of GIT1/βPIX assembly and Rac1 signaling by GluN3A. (A) Extracts from cortical neurons infected with Sindbis virus expressing GFP (control), GFP-GluN3A, or GFP-GluN3AΔGIT1 were immunoprecipitated with GIT1 antibody or control IgG and probed for GIT1 and βPIX. Representative blots and quantification are shown. Bound βPIX levels were normalized to GIT1 levels in the IP (n = 5 independent experiments; *P < 0.05, ANOVA followed by Tukey’s test). Here, and in all subsequent figures, error bars indicate the mean ± SEM. (B) Inhibition of Rac1 activation in neurons expressing GFP-GluN3A (n = 5 independent experiments; ***P < 0.001, ANOVA followed by Tukey’s test). (C) Diagram depicts the effects of GluN3A on GIT1 function. GIT1 forms a complex with the Rac1-GEF βPIX that promotes the exchange of Rac1-bound GDP for GTP; the assembly of this complex is disrupted by GluN3A (Left) but not by GluN3A lacking the GIT1-binding domain (Right). (D) Forebrain extracts from P8 WT (wt) and GluN3A^−/− mice were immunoprecipitated with GIT1 antibodies and blotted for GIT1 and βPIX. Representative blots and quantification are shown (n = 9 independent experiments; *P < 0.05, Student’s t test). (E) Increased Rac1-GTP and pPAK levels in extracts from P8 GluN3A^−/− mouse forebrain without changes in other NMDAR-dependent signaling pathways (n = 6–10 independent experiments; *P < 0.05 and **P < 0.01, Student’s t test).

Fig. 3.

GluN3A alters the targeting of GIT1 to synapses. (A and B) Biochemical fractionation analysis of WT and GluN3A^−/− mouse forebrain at P8. Shown are representative IBs of whole homogenates (H) and TIFs probed with the indicated antibodies (A) and quantification (B) (n = 7–9 independent experiments; *P < 0.05, Student’s t test). Plotted are protein levels in TIFs normalized to levels in homogenates. (C) Representative single confocal images of Homer and GIT1 immunostaining in GluN3A^−/− cultured hippocampal neurons infected with Sindbis virus expressing GFP (control), GFP-GluN3A, or GFP-GluN3AΔGIT1. (Scale bar, 1 μm.) (D) Quantification of the effects of GFP, GFP-GluN3A, or GFP-GluN3AΔGIT1 on GIT1 cluster size (n = 11–13 neurons from four independent cultures; P < 0.001, Kolmogorov–Smirnov test) and ratio of GIT1 clustered/nonclustered fluorescence intensities (n = 11–13 neurons from four independent cultures; ***P < 0.001, ANOVA followed by Tukey’s test). (E) Colocalization with the postsynaptic marker Homer measured using Pearson’s coefficient (n = 11–13 neurons from four independent cultures; ***P < 0.001 ANOVA followed by Tukey’s test).

Fig. 4.

Synaptic activity regulates GluN3A binding to GIT1. (A) Representative images of cultured hippocampal neurons, either untreated (control) or exposed to TTX or bicuculline for 45 min and stained with the indicated antibodies. (Scale bar, 1 μm.) (B) Quantification of the ratios of GIT1 and βPIX clustered/nonclustered fluorescence intensities (n = 10–13 neurons from three independent cultures; *P < 0.05, **P < 0.01, and ***P < 0.001, ANOVA followed by Tukey’s test). Bic, bicuculline. (C) Cultured neurons treated with TTX or bicuculline were lysed, and lysates were immunoprecipitated with GluN3A or GIT1 antibodies or control IgG. Inputs (10% of lysate) and IPs were blotted with the indicated antibodies. (D) Quantification of GIT1/GluN3A and GIT1/βPIX binding. Bound GIT1 levels were normalized to GluN3A levels in the IP, and bound βPIX was normalized to GIT1 levels in the IP (n = 5 independent experiments; *P < 0.05, **P < 0.01, and ***P < 0.001, ANOVA followed by Tukey’s test). (E) Representative IBs of cortical neurons pretreated with TTX and stimulated with 50 μM glutamate for 5 min in the presence or absence of 10 μM CNQX or 50 μM APV. Lysates were immunoprecipitated with GluN3A and immunoblotted with the indicated antibodies. (F) Quantification of GIT1/GluN3A binding (n = 6 independent experiments; **P < 0.01 and ***P < 0.001, ANOVA followed by Tukey’s test). (G) Model depicts the potential role of activity-dependent regulation of GluN3A/GIT1 binding on spine maturation. During postnatal development, GluN3A limits the maturation of inactive synapses by preventing the localization of GIT1 to synapses via mechanisms that might involve a limited anchoring of GluN3A itself to postsynaptic compartments (arrows) or by inhibiting GIT1 interactions with βPIX. Synaptic activity releases GIT1 from GluN3A, allowing its synaptic localization and the recruitment of actin regulators. Local activation of Rac1 promotes actin cytoskeleton remodeling and, in turn, spine growth and stabilization.

Fig. 5.

GIT1 binding is required for the negative regulatory effects of GluN3A on spine morphogenesis. (A) Representative confocal images of cultured hippocampal neurons cotransfected with GFP (control), GFP-GluN3A, or GFP-GluN3AΔGIT1, as well as mCherry at DIV10, and fixed at DIV17. (Scale bar, 2 μm.) (B) Quantitative analysis of total spine densities or spine densities across different morphological categories (secondary and tertiary dendrites of n = 16–18 neurons from three different cultures; **P < 0.01 and ***P < 0.001, ANOVA followed by Tukey’s test). (C) Morphometric analysis of spine head diameter (n = 750–1,600 spines; P < 0.005, Kolmogorov–Smirnov test).