The RacGAP β2-Chimaerin selectively mediates axonal pruning in the hippocampus

Abstract

Axon pruning and synapse elimination promote neural connectivity and synaptic plasticity. Stereotyped pruning of axons that originate in the hippocampal dentate gyrus (DG) and extend along the infrapyramidal tract (IPT) occurs during postnatal murine development by neurite retraction and resembles axon repulsion. The chemorepellent Sema3F is required for IPT axon pruning, dendritic spine remodeling, and repulsion of DG axons. The signaling events that regulate IPT axon pruning are not known. We find that inhibition of the small G protein Rac1 by the Rac GTPase-activating protein (GAP) β2-Chimaerin (β2Chn) mediates Sema3F-dependent pruning. The Sema3F receptor neuropilin-2 selectively binds β2Chn, and ligand engagement activates this GAP to ultimately restrain Rac1-dependent effects on cytoskeletal reorganization. β2Chn is necessary for axon pruning both in vitro and in vivo, but it is dispensable for axon repulsion and spine remodeling. Therefore, a Npn2/β2Chn/Rac1 signaling axis distinguishes DG axon pruning from the effects of Sema3F on repulsion and dendritic spine remodeling.

Figure 1. Down-regulation of Rac-GTP Levels is Required for Infrapyramidal Tract Pruning

(A and B) Immunostaining of DIV14 hippocampal neurons treated with 10nM AP-control (A) or Sema3F-AP (B), using GST-PAK1 (top panels, green; middle panels, white) and Tau1 (top panels, red). Bottom panels: black and white mask showing only GST-PAK1⁺ puncta in Tau1⁺ axons. Scale bar, 5µm. (C) Quantification of Rac/Cdc42-GTP puncta in hippocampal neurons presented as puncta per µm² of axonal fluorescence in AP-treated (0.24±0.007 puncta/µm²) and Sema3F-AP treated cultures (0.121±0.0015 puncta/µm²; two tailed t-test, n=3 experiments, ***p= 0.00082). (D–F) Expression of control EGFP (D and F) or RacQL-ires-EGFP (E) from lentivirus stereotactically injected into the dentate gyrus at P20 and shown here at P45. Coronal brain sections were immunostained for calbindin (middle panel in white, top panel red), GFP (bottom panel in white, top panel in green) and counterstained with ToproIII (top panel, blue). (E) The IPT is not pruned in the presence of constitutively active Rac. Scale bar, 100 µm. (F) GFP control lentivirus injection showing a representative region of interest magnified in (D) and (E). Red arrowheads mark the distal end of calbindin⁺/GFP⁺ IPT fibers. Scale bar, 250 µm. (G) Quantification of the ratio of GFP⁺, calbindin⁺ IPT length to the length of the MB in control (0.469±0.07, n=7) and RacQL-IRES-EGFP injected animals (0.782±0.115, n=6, two-tailed t-test: ***p= 0.00011; error bars, s.d.). (H and I) Immunostaining for vGlut1 (bottom panel white, top panel blue), calbindin (top panel, red) and GFP (top panel, green) in EGFP-expressing (H) or RacQL-ires-EGFP–expressing (I) lentivirus-injected DG granule cell axons extending within the IPT. Yellow arrows mark presynaptic vGlut1⁺ terminals present in RacQL-ires-EGFP–expressing distal IPT axons (I). Scale bar, 50 µm.

Figure 2. The Rac-GAP β2-Chimaerin Binds to Neuropilin-2 and is Activated at the Membrane by Sema3F

(A) Co-immunoprecipitation of GFP tagged chimaerins (α2Chn or β2Chn) with Myc tagged neuropilin-1 (Npn-1), neuropilin-2 (Npn-2) or plexinA3 (PlexA3) in HEK293 cells in vitro. Npn-1, Npn-2 and PlexA3 were immunoprecipitated with antibodies directed against the Myc tag and co-precipitation of chimaerins was detected. Note the strong interaction between β2Chn and Npn-2. (B) Schematic of FRET probes used to evaluate β2Chn activation (Wang et al., 2006). (C and D) Analysis of the β2Chn–Rac1 interaction by FRET at the membrane (full circles) or in the cytoplasm (open circles) of Neuro2A cells treated with 10 nM AP (C) or Sema3F (D). FRET was measured every 6 s after ligand treatment for 10 min. In the presence of Sema3F, a dramatic increase in FRET is observed selectively at the cell membrane, revealing that β2Chn is activated as evidenced by its binding to Rac1 and recruitment to the cell membrane. Error bars, s.e.m. (E) Quantitative analysis of maximum FRET in the peripheral region after AP or Sema3F treatment. Average Max ΔFRET: AP:0.0096±0.013; Sema3F: 0.0944±0.0141; t-test ***p<0.0001, n=34–37 membrane regions per treatment. Data are expressed as mean±s.e.m. (F) CoIP of endogenous Npn-2 and GFP-tagged β2Chn in Neuro2A cells in the presence of different Sema3F concentrations. Npn-2 was immunoprecipitated with anti-Npn-2 and GFP-β2Chn was detected using anti-GFP. Bath application of 5nM and 10nM Sema3F for 20 minutes causes a significant, dose-dependent, reduction in β2Chn binding to Npn-2. (G) Postnatal expression of βChn in the dentate gyrus assessed by in situ hybridization. βChn DG levels progressively increase after birth and peak during IPT pruning (P30–P45). Scale bar, 100 µm.

Figure 3. β2-Chimaerin is Required for Sema3F-dependent Pruning, but not Repulsion, in vitro

(A–D) WT (A and B) and Chn2^−/− (C and D) hippocampal neurons were treated with 10nM AP (control) (A and C) or Sema3F-AP (B and D) and immunostained with GST-PAK1 (top panels, green) and with anti-Tau1 (top panels, red). Bottom panels, mask for GST-PAK1 positive puncta in axons. Scale bar, 5 µm. (E) Quantification of active Rac/Cdc42 in axons treated with AP or Sema3F expressed as a percentage of GST-PAK⁺ puncta/µm² in AP-treated WT neurons (n=3, WT-AP:100%; WT-Sema3F: 52.93±3.63%; Chn2^−/−-AP: 95.95±7.51%; Chn2^−/−-Sema3F: 100.01±4.49%; ANOVA, p<0.0001, Tukey HSD Test **p<0.01. error bars, s.d.). (F–I) Synaptoporin (SOP, green) and Tau1 (red) immunolabelling of WT (F and G) and Chn2^−/− (H and I) DIV21 hippocampal neurons treated with AP (F and H) or Sema3F (G and I). SOP levels are notably reduced following bath application of Sema3F to WT neurons (G), but this response is abolished in Chn2^−/− neurons (I). Scale bar, 10 µm. (J) Quantification of SOP labeling assay (n=4 experiments, 10 fields per experiment, ANOVA p=0.000927, followed by Tukey HSD test, *p<0.05 compared to all other treatments; error bars, s.e.m.). WT-AP: 0.0511±0.0071; WT-Sema3F: 0.0238±0.0035; Chn2^−/−-AP: 0.0537±0.0062; Chn2^−/−-Sema3F: 0.0535 ± 0.0054 SOP puncta/µm². (K–N) Dentate gyrus explants from P2 WT (K and L) or Chn2^−/− (M and N) mice were used in stripe assays with alternating AP stripes (K and M) or alternating AP and Sema3F stripes (L and N). Both WT and Chn2^−/− neurites steer away from Sema3F stripes. n=16–20 explants per treatment, from 4–5 animals per genotype. Scale bar, 100 µm. (O) Quantification of neurite outgrowth performed on WT and Chn2^−/− dentate gyrus explants (see Figure S4). n.s., not significant. error bars, s.e.m.

Figure 4. β2-Chimaerin is Required for IPT Pruning in vivo, but is Dispensable for DG Dendritic Spine Remodeling

(A–D) Immunostaining of WT (A and C) and Chn2^−/− (B and D) P45 hippocampi with anti-calbindin (red), anti-vGlut1(green) and ToproIII (blue). The IPT is notably longer in Chn2^−/− mice (B) compared to WT (A). Yellow arrows mark the distal end of the IPT. Scale bar, 100 µm. (C and D) Higher magnification views of (A) and (B), respectively. vGlut1⁺ presynaptic terminals are observed in the distal region of the IPT in Chn2^−/− mice (D). Scale bar, 50 µm. (E) Quantification of IPT pruning, expressed as the ratio of IPT length to the length of the MB in CA3. The IPT is significantly longer in Chn2^−/− mice (0.87±0.034, n=12 brain hemispheres from 8 mutant mice) than in WT (0.515±0.046, n=13 hemispheres from 8 WT mice, two tailed t-test ***p=8.02*10⁻¹⁷; error bars ± s.d). (F and G) Transmission electron micrographs of distal IPT regions in WT (F) and Chn2^−/− (G) mice. Insets show a single axon terminal and PSD in WT (F), and a characteristic mossy fiber asymmetric synapse with two PSDs in Chn2^−/− (G) mice. Asterisks mark PSDs. Scale bar, 500nm for F and G, and 150nm for insets. (H) Quantification of synapses in the distal infrapyramidal region of WT and Chn2^−/− mice determined from EM analysis. WT: 0.042±0.003 psd/µm², Chn2^−/−: 0.265±0.069 psd/µm², two tailed t-test, t-test **p=0.005, error bars=s.e.m. (I and J) Chn2 is not required for DG granule cell dendritic spine or anterior commissure development. Golgi staining of adult DG granule cell dendrites in WT (I) and Chn2^−/− (J) hippocampi. Scale bar, 10 µm for I and J and 6 µm for insets. (K) Quantification of WT and Chn2^−/− DG granule cell dendritic spine density; no significant difference between WT and Chn2^−/− is observed (t-test: p=0.83 for 0–25 µm, p=0.11 for 50–75 µm; error bars ± s.d.).

Figure 5. A Hyperactive Form of β-Chimaerin is Sufficient for IPT Pruning in vivo

(A and B) Immunohistochemistry with anti-calbindin (red) and ToproIII (blue) on WT and Chn2^I130A/I130A P28 hippocampi. Scale bar, 100 µm. (C) Quantification of the IPT length-to-MB length ratio in WT (0.664±0.057) and Chn2^I130A/I130A (KI/KI) mice (0.495±0.072, n=8, two-tailed t-test **p=0.00596). (D and E) Genetic interaction between Npn-2 and Chn2. (D) Immunostaining using anti-calbindin (green) and ToproIII (blue) of WT (top left), Npn-2^+/− (top right), Chn2^+/− (bottom left) and Chn2^+/−; Npn2^+/− transheterozygotes (bottom right) P45 hippocampi. Scale bar, 100 µm. (E) Quantification of the genetic interactions between Chn2 and Npn2 (n=7 hemispheres from 5 animals for each genotype, ANOVA p<0.0001; Tukey HSD test **p<0.01 compared to all other genotypes). WT: 0.45±0.053; Chn2^+/−: 0.463±0.046 ; Npn-2^+/−: 0.447±0.024; Chn2^+/−; Npn2^+/− :0.698±0.092. Yellow brackets delineate IPT length in A, B, and D. error bars are s.d.

Figure 6. β2Chn is Required in the Dentate Gyrus for IPT Pruning in vivo. (A–C)

Histological analysis of hippocampi from control (A), shRNA2 (B), shRNA4 (C) lentivirus-injected animals using anti-GFP (top panels, green; bottom panels, white), anticalbindin (top panels, red; middle panels white) and ToproIII (top panels, blue). The IPT in shRNA2- and shRNA4-injected animals (B and C) extends almost as far as the distal blade of the MB, as compared to control-injected animals in which the IPT extends only 45% of the MB length (A). Arrowheads mark the distal end of the calbindin⁺, GFP⁺ IPT fibers. Scale bar, 100µm. (D–F) β2-chimaerin is required in the dentate gyrus for IPT presynaptic pruning. Immunostaining for vGlut1 (bottom panels, white; top panel, blue), calbindin (top panels, red) and GFP (middle panels, white; top panels, green) of control (D), shRNA2 (E), shRNA4 (F) lentivirus-injected mice. Ectopic vGlut1⁺ presynaptic terminals are present in the distal region of the IPT in shRNA injected hippocampi (E and F). Arrows point to presynaptic vGut1⁺ terminals present in shRNA injected EGFP⁺ distal IPT axons. Scale bar, 50 µm.

Figure 7. β2Chn GAP Function is Required in the Dentate Gyrus for IPT Pruning in vivo

(A and B) Immunostaining of hippocampal sections obtained from shRNA2+WT* (A), and shRNA2+ΔEIE* (B) lentivirus-injected animals using anti-GFP (top panels, green; bottom panels, white), anti-calbindin (top panels, red; middle panels white) and ToproIII (top panels, blue). The defect observed in shRNA injected animals (see Figure 6) can be rescued by human WT β2Chn (A), but not by a human β2Chn harboring a three amino acid deletion rendering the GAP domain inactive (ΔEIE, B). Arrowheads mark the distal end of the calbindin⁺, GFP⁺ IPT fibers. Scale bar, 100µm. (C) Quantification of GFP⁺, calbindin⁺ IPT length expressed as a ratio of IPT/MB length. Control: 0.47±0.07, n=7; shRNA2: 0.78±0.013, n=6; shRNA4: 0.83±0.07, n=5; shRNA2+WT*: 0.43±0.06, n=9; shRNA2+ΔEIE*: 0.77±0.04, n=6. ANOVA (p<0.0001) followed by Tukey HSD test, **p<0.01 compared to control and shRNA2+WT*; error bars ± s.d. (D and E) The Rac-GAP activity of β2Chn is required in the dentate gyrus for IPT presynaptic pruning. Immunohistochemistry for vGlut1 (bottom panels, white; top panel, blue), calbindin (top panels, red) and GFP (middle panels, white; top panels, green) of shRNA2+WT* (D), and shRNA2+ΔEIE* (E) lentivirus-injected mice. Injection of human shRNA-resistant WT β2̃Chimaerin rescues the accumulation of vGlut1 in the distal IPT region (D), however a human GAP-deficient form of β2̃Chimaerin (ΔEIE) (E), fails to rescue the IPT pruning defect observed in shRNA-injected hippocampi. Arrows point to presynaptic vGut1⁺ terminals present in shRNA injected EGFP⁺ distal IPT axons (E). Scale bar, 50 µm.