The schizophrenia risk gene C4 induces pathological synaptic loss by impairing AMPAR trafficking

Abstract

Neuroimmune interactions play a significant role in regulating synaptic plasticity in both the healthy and diseased brain. The complement pathway, an extracellular proteolytic cascade, exemplifies these interactions. Its activation triggers microglia-dependent synaptic elimination via the complement receptor 3 (CR3). Current models of pathological complement activity in the brain propose that accelerated synaptic loss resulting from overexpression of C4 (C4-OE), a gene associated with schizophrenia, follows this pathway. Here, we report that C4-mediated cortical hypoconnectivity is CR3-independent. Instead, C4-OE triggers impaired GluR1 trafficking through an intracellular mechanism involving the endosomal protein SNX27, resulting in pathological synaptic loss. Moreover, C4 circuit alterations in the prefrontal cortex, a brain region associated with neuropsychiatric disorders, were rescued by increasing neuronal levels of SNX27, which we identify as an interacting partner of this neuroimmune protein. Our results link excessive complement activity to an intracellular endo-lysosomal trafficking pathway altering synaptic plasticity.

Fig. 1. C4 overexpression led to mPFC pyramidal neuron hypoconnectivity through a CR3-independent mechanism.

A Left: In utero electroporation (IUE) procedure performed on E16 embryos. GFP-control (Con) represents the transfection of a single plasmid (GFP under the CAG promoter), while C4-OE represents the transfection of two plasmids (GFP and C4, both under the CAG promoter). Right: Representative 60X confocal image of P21-23 L2/3 mPFC neurons transfected with GFP (green). DAPI, cytoarchitecture. White dotted line, pia. Scale bar = 100 μm. B The model depicts the mechanism of spine removal through C3b recognition by the microglia expressing CR3 [54, 77]. C Left: Representative confocal image (60X) of CR3KO-control and CR3KO C4-OE. GFP, white signal. Scale bar = 2 μm. Right: C4-OE in mice lacking the CR3 (red circles) showed decreased dendritic spine density relative to control CR3KO mice (blue squares). t-test. *p < 0.05. D Structure of wt C4 and C4 mutants (C4-ΔC345C and C4-GFP). C4 comprises three main chains [27] (β, α and γ) and a highly conserved C-terminal domain known as C345C [27] (cyan). E Left: Representative confocal image (60X) of Control and C4 mutants. Right: OE of wt C4 and mutants C4-GFP-OE and ΔC345C-OE caused a decrease in dendritic spine density relative to GFP-control. Cells were filled with mRuby3 instead of GFP for analysis of the C4-GFP-OE condition. One-way ANOVA. **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar = 2 μm. C N = 13 CR3KO-control dendrites, 3 animals; N = 12 CR3KO-C4-OE dendrites, 3 animals, E N = 9 GFP-control dendrites, N = 10 C4-OE and C4-ΔC345C-OE dendrites, N = 11 C4-GFP dendrites, 3 animals each. All graphs, Mean ± SEM.

Fig. 2. SNX27 interacts with C4 and is essential for normal dendritic spine formation of L2/3 PFC neurons.

A Western blot (WB) showing co-IP from HEK293T cells transfected with SNX27 and either C4, C4-ΔC345C or C4-GFP plasmids. The presence of proteins was detected by Western blotting in cell lysates (Input). Interaction of wt C4 and C4 mutants with SNX27 was detected by Western blotting of co-IP against FLAG-tagged SNX27 (IP: FLAG). Protein molecular weights, kilodaltons (kDa). + indicated the presence of wild type protein. B Left: Representative confocal image (60X) of scr-shRNA control (solid blue triangles), shSNX27 (solid red triangles), C4-OE + shSnx27 (orange open circles) and C4-OE (red open circles). GFP, white signal. Scale bar = 2 μm. Right: SNX27 KD and C4-OE + SNX27 KD led to a decrease in spine density relative to scr-shRNA to the same levels as C4-OE. One-way ANOVA. *p < 0.05, **p < 0.01. C Representative STED images (60X) showing GFP-positive dendritic spine (white), SNX27 (green) and C4 (magenta) of DIV 14 cultured PFC neurons. White dotted line indicates spine morphology. Scale bar = 0.5 μm. D Normalized intensity profile of a line drawn through a spine head with SNX27 and C4 colocalized clusters in (A) showing intensities of GFP (gray line), SNX27 (green) and C4 (magenta). E Orthogonal views of (C) are shown (XY, YZ and XZ). Yellow arrowhead, SNX27 and C4 colocalization in the spine. F Representative STED images (60X). White dotted line, dendritic morphology. Scale bar = 2 μm. G Analysis C4 and SNX27 signals showed that approximately 41% of C4 area colocalized with SNX27 and approximately 44% of SNX27 area colocalized with C4 in spines and dendrites. H Representative confocal image (40X) of C4-GFP signal in electroporated L2/3 neurons. Scale bar = 15 μm. Boxes outlined by magenta and yellow show zoomed-in views of the insets, respectively. Scale bar = 5 μm. I 3D surface reconstructions of the insets in magenta and yellow, respectively, from (H). Scale bar = 2 μm. B N = 10 scr-shRNA control, shSnx27 and C4-OE + shSnx27 dendrites, N = 11 C4-OE dendrites, 3 animals each. All graphs, Mean ± SEM. F, G N = 7 dendritic segments, from 3 cells. All graphs, Mean ± SEM.

Fig. 3. Increasing neuronal levels of SNX27 rescued the C4 hypoconnectivity phenotype.

A Representative confocal image (60X) of GFP-positive P21–23 L1 apical tufts in GFP-control (Con, blue squares), C4-OE (red circles), C4-SNX27-OE (purple squares) and SNX27-OE (gray square) in the mPFC. GFP, white signal. Scale bar = 2 μm. B C4 synaptic pathology was rescued by simultaneous OE of SNX27 (C4-SNX27-OE). One-way ANOVA. *p < 0.05, **p < 0.01. C Representative traces of mEPSCs in GFP-control (blue), C4-OE (red), C4-SNX27-OE, (purple), and SNX27-OE (gray) L2/3 mPFC pyramidal neurons from P20–25 animals. Scale bars, 125 ms,10 pA. D mEPSC frequency was decreased in C4-OE relative to GFP-control (Kruskal–Wallis one-way ANOVA. **p < 0.01; GFP-control vs. C4-OE, Dunn’s multiple comparison test. ***p < 0.001) but was rescued when C4 and SNX27 were co-overexpressed (C4-OE vs. C4-SNX27-OE, Dunn’s multiple comparison test. *p < 0.05). SNX27-OE alone did not alter mEPSC frequency relative to GFP-control (Con vs. SNX27-OE, Dunn’s multiple comparison test. p > 0.9999) but was significantly greater than in C4-OE (C4-OE vs. SNX27-OE, Dunn’s multiple comparison test. *p < 0.05). Hz Hertz. There were no differences in the amplitude (E, one-way ANOVA. p = 0.6241), Rise 10–90 (F, Kruskal–Wallis one-way ANOVA. p = 0.4346), and Decay tau (G, one-way ANOVA. p = 0.2122). N = 24 GFP-control cells, from 6 animals. N = 20 C4-OE cells, from 4 animals. B N = 8 control dendrites, N = 9 C4-SNX27-OE dendrites, N = 11 C4-OE and SNX27-OE dendrites, 3 animals each. D–G N = 25 C4-SNX27-OE cells, from 4 animals. N = 22 SNX27-OE cells, from 5 animals. All graphs, Mean ± SEM.

Fig. 4. C4-OE disrupted GluR1-containing AMPAR trafficking in L1 apical tuft dendritic spines.

A Model depicting effects of C4 overexpression on GluR1 recycling in dendritic spines. B Representative images (60X) showing a GFP-positive dendritic spine (white), GluR1 (green) and Rab11a (magenta) of P21–23 apical tufts in GFP-controls (blue frame). C Representative images (60X) showing a dendritic spine identified with GFP signal (white), GluR1 (green) and Rab11a (magenta) in C4-OE (red frame). B, C Yellow arrowhead, GluR1 or Rab11a clusters in spines. White-filled arrowhead, GluR1/Rab11a colocalization. White empty arrowhead, non-colocalized GluR1/Rab11a. Spine silhouette, white dotted line. Orthogonal views are shown (XY, YZ and XZ). Scale bar = 2 μm. D C4-OE caused a 47% decrease in the amount of GluR1 colocalized with Rab11a compared to GFP-control. E There was no change in the amount of Rab11a colocalized with GluR1 in C4-OE relative to GFP-control. F C4-OE increased the minimum distance between GluR1 and Rab11a clusters by 35% relative to GFP-control. Green circle: GluR1, Magenta circle: Rab11a. G C4-OE led to a 25% decrease in the overlapping volume between GluR1 and Rab11a relative to GFP-control. Green circle: GluR1, Magenta circle: Rab11a. H Schematic showing the effects of C4-OE on GluR1 degradation in dendritic spines. I Representative images (60X) showing a GFP-positive spine (white), GluR1 (green) and LAMP1 (magenta) in GFP-control (blue frame). J Representative images (60X) showing a GFP-positive spine (white), GluR1 (green) and LAMP1 (magenta) in C4-OE (red frame). I, J Yellow arrowhead, GluR1 or Rab11a clusters in spines. White empty arrowhead, non-colocalized GluR1/LAMP1. White-filled arrowhead, GluR1/LAMP1 colocalization. Spine silhouette, white dotted line. Orthogonal views are shown (XY, YZ and XZ). Scale bar = 2 μm. K C4-OE led to a 145% increase in the amount of GluR1 colocalized with LAMP1. L C4-OE caused a 103% increase in the amount of LAMP1 colocalized with GluR1. M C4-OE induced a 31% decrease in the minimum distance between GluR1 and LAMP1 clusters. Green circle: GluR1, Magenta circle: LAMP1. N Compared to GFP-control, C4-OE did not alter the overlapping volume between GluR1 and LAMP1. Green circle: GluR1, Magenta circle: LAMP1. D, E, K, L N = 8 dendrites, 3 animals for Con and C4-OE. F, G, M, N N = 7 dendrites, 3 animals, Con; and N = 8 dendrites, 3 animals, C4-OE. D–G, K–N t-test. *p < 0.05, **p < 0.01, ***p < 0.001. All graphs, Mean ± SEM.

Fig. 5. C4-OE led to increased levels of GluR1-containing AMPAR in mPFC neurons.

A Left: Schematic of dissected GFP-positive tissue from IUE animals, used for isolating cytosolic and synaptosome fractions. Right: Western blot (WB) showing levels of GluR1 with C4 overexpression. GluR1 levels were detected in synaptosome and cytosolic fractions in control and C4-OE conditions. Since C4 was expressed at relatively low levels, it could only be detected when several brains were pooled. Therefore, each lane corresponds to pooled brain lysate from a single litter (6 mice). Protein molecular weights, kilodaltons (kDa, left column). “+” indicates the presence of C4 (top row). Numbers in the bottom row indicate fold change in GluR1 levels relative to control synaptosome and cytosolic fraction. B Left: Schematic of degradation assay in HEK293T cells. Right: WB quantified protein levels of GluR1 in GS and GSC cells with CHX and CHX + MG132 treatment. Protein molecular weights, kilodaltons (kDa, left column). “+” indicates the presence of C4, CHX or MG132. C Left: CHX treatment led to decreased GluR1 levels in GS cells, which was rescued upon application of MG132. Right: GluR1 levels in GSC cells were unaffected by CHX or CHX + MG132 treatments. Light gray circles: DMSO treated cells, Dark gray triangles: CHX treated cells, Dark gray squares: CHX + MG132 treated cells. N = 3 sample replicates per condition. One-way ANOVA. *p < 0.05, **p < 0.01. All graphs, Mean ± SEM.