Narp regulates homeostatic scaling of excitatory synapses on parvalbumin-expressing interneurons

Abstract

Homeostatic synaptic scaling alters the strength of synapses to compensate for prolonged changes in network activity and involves both excitatory and inhibitory neurons. The immediate-early gene Narp (neuronal activity-regulated pentraxin) encodes a secreted synaptic protein that can bind to and induce clustering of AMPA receptors (AMPARs). We found that Narp prominently accumulated at excitatory synapses on parvalbumin-expressing interneurons (PV-INs). Increasing network activity resulted in a homeostatic increase of excitatory synaptic strength onto PV-INs that increased inhibitory drive and this response was absent in neurons cultured from Narp-/- mice. Activity-dependent changes in the strength of excitatory inputs on PV-INs in acute hippocampal slices were also dependent on Narp and Narp-/- mice had increased sensitivity to kindling-induced seizures. We propose that Narp recruits AMPARs at excitatory synapses onto PV-INs to rebalance network excitation/inhibition dynamics following episodes of increased circuit activity.

Figure 1

Narp expression is highly enriched at excitatory synapses on PV-INs. (a) Representative image of hippocampal neuronal cultures stained with Narp (green) and the neuronal dendritic marker MAP2 (red). Inset: dendrite from a neuron with very little detectable surface Narp (purple border) and a dendrite from a neuron with an accumulation of surface Narp (blue border). Scale bars represent 100 μm and 5 μm (inset) (b) Cultured hippocampal neurons live-labeled with an antibody against Narp prior to immunostaining against cell type-specific markers PV (left), Calretinin (middle), or CAMKIIα (right). Scale bars represent 10 μm. (c) Summary of the data shown in b. Narp intensity per μm dendrite for each cell type was normalized to PV expressing neurons (PV, 100% ± 19.41%, n = 14 cells; Calretinin, 11.84% ± 3.00%, n = 15 cells; CAMKIIα, 9.72% ± 2.36%, n = 15 cells). Statistical analysis was performed using a nonparametric one-way ANOVA test. ** P < 0.01 vs Parvalbumin group. Error bars represent s.e.m. (d) Immunohistochemical staining for PV (red) and Narp (green) in the CA3 region of the hippocampus. Scale bar represents 100 μm. (e) Cultured hippocampal neurons were live-labeled with Narp antibody prior to immunostaining against PV and the excitatory post-synaptic marker PSD95. Arrows indicate co-localized punctae. Scale bars represent 10 μm.

Figure 2

Narp expression on PV-INs is dynamically regulated by activity (a) Following treatment for 48 hours with either 1 μM TTX (middle), control (left), or 40 μM bicuculline (right), cultured neurons were immunostained for PV and surface Narp. Scale bars represent 10 μm (top) and 5 μm (bottom). (b) Time course of the data shown in a. Narp intensity per μm dendrite after bicuculline (left) or TTX (right) treatment was normalized to 0 hour (untreated) group (Bicuculline: 0h, 100% ± 8.54%, n = 35 cells; 4h, 187.4% ± 45.47%, n = 15 cells; 12h, 241.4% ± 85.93%, n = 15 cells; 24h, 653.6% ± 126.5%, n = 15 cells; 48h, 2,770% ± 633.2%, n = 35 cells. TTX: 0h, 100% ± 10.61%, n = 35 cells; 4h, 52.27% ± 9.51%, n = 15 cells; 12h, 22.00% ± 4.58%, n = 15 cells; 24h, 18.03% ± 2.43%, n = 15 cells; 48h, 46.96% ± 6.51%, n = 32 cells) Statistical analysis was performed using a nonparametric one-way ANOVA test. ** P < 0.01 and *** P < 0.001 vs 0h group. Error bars represent s.e.m. (c) Representative western blot showing levels of surface Narp, Transferrin receptor (TfR), and Actin levels in untreated control cultures (center) and after 48 hour treatment with TTX (left) or bicuculline (right). Full-length blots are presented in Supplementary Figure 2. (d) Summary of the data shown in c. All values are presented as a ratio of surface Narp intensity/surface TfR intensity and were normalized to untreated control (Untreated, 100%, n = 3; 1 μM TTX, 26.15% ± 15.73%, n = 3; 40 μM Bicuculline, 110% ± 10.32%, n = 3). Statistical analysis was performed using a nonparametric one-way ANOVA test. ** P < 0.01 as indicated by bracket. Error bars represent s.e.m. (e) Summary of Narp RT-PCR. All Narp mRNA values are normalized to paired GAPDH mRNA values and the grouped Narp mRNA averages for each treatment are normalized to untreated control cultures and presented as a fold difference. (Untreated, 1, n = 3; 1 μM TTX, 0.57 ± 0.25, n = 3; 40 μM Bicuculline, 9.12 ± 5.18, n = 3). Statistical analysis was performed using a repeated measures ANOVA test. * P < 0.05. Error bars represent s.e.m.

Figure 3

Narp is derived from presynaptic neurons contacting PV-INs (a) Unlabeled WT hippocampal neurons plated with a small population of CM-DiI-labeled Narp^−/− neurons were stained for surface Narp and Parvalbumin. (b) Same preparation as in a, but with labeled WT and unlabeled Narp^−/− neurons. (c) Summary of the data shown in a and b. Narp intensity per μm dendrite for each condition was normalized to unlabeled WT (WT pre/WT post, 100% ± 13.65%, n = 15 cells; ^−/− pre/^−/− post, 41.85% ± 22.53%, n = 14 cells; WT pre/^−/− post, 155% ± 28.5%, n = 15 cells; ^−/− pre/WT post, 21.56% ± 4.76%, n = 15 cells). Statistical analysis was performed using a nonparametric one-way ANOVA test. ** P < 0.01 vs WT pre/WT post group.

Figure 4

Disruption of perineuronal nets results in loss of surface Narp accumulation. (a) (top) Untreated cultured hippocampal neurons stained for surface Narp and PV. (bottom) Representative Narp and PV immunostaining after 48 hour treatment with 0.2 U Chondroitinase ABC. Scale bars represent 10 μm. (b) Summary of the results seen in a. Narp intensity per μm dendrite for each condition was normalized to untreated control (Untreated, 100% ± 7.27%, n = 20 cells; Chondroitinase ABC, 45.83% ± 4.92%, n = 20 cells, *** P < 0.001 vs Untreated group; Bicuculline + Chondroitinase ABC 48 hours, 52.47% ± 4.83%, n = 19 cells). ** P < 0.01 vs Untreated group. Error bars represent s.e.m. (c) Representative western blot showing levels of surface Narp and Transferrin receptor (TfR) levels in untreated control cultures (left) and after 48 hour treatment with 0.2 U ChABC (center) or 0.2 U ChABC and 40 μM bicuculline (right). The full-length blot is presented in Supplementary Figure 2. (d) Summary of the data shown in c. All values are presented as a ratio of surface Narp intensity/surface TfR intensity and were normalized to untreated control.(Untreated, 100%, n = 3; 0.2 U ChABC, 96.45% ± 8.18%, n = 3; 0.2 U ChABC + 40 μM Bicuculline, 210% ± 34.37%, n = 3). Statistical analysis was performed using a nonparametric one-way ANOVA test. ** P < 0.01 vs. Untreated group. Error bars represent s.e.m.

Figure 5

Narp modulates GluR4 levels on PV-INs in an activity-dependent manner. (a) Representative GluR4 levels on cultured WT PV-INs in the presence of TTX (right), no treatment (middle), or bicuculline (right). Scale bars represent 10 μm (top) and 5 μm (inset). (b) Summary of the data shown in a and c. GluR4 intensity per μm dendrite for all treatments were normalized to WT untreated neurons (WT Untreated, 100% ± 17.71%, n = 17 cells; WT TTX, 35.10% ± 7.73%, n = 20 cells; WT Bicuculline, 229.19% ± 36.67%, n = 20 cells; Narp^−/− Untreated, 45.70% ± 7.76%, n = 20 cells; Narp^−/− TTX, 65.93% ± 13.24%, n = 20 cells; Narp^−/− Bicuculline, 20.49% ± 6.47%, n = 20 cells) Statistical analysis was performed using a nonparametric one-way ANOVA test. *** P < 0.001 vs Untreated WT group or as indicated by bracket. Error bars represent s.e.m. (c) Same experiment as in a except with Narp^−/− PV-INs. (d) Time course of PV-IN GluR4 levels during TTX (white bars) or bicuculline (black bars) treatment. GluR4 intensity per μm dendrite was normalized to 0 hour (untreated) group (Bicuculline: 0 h, 100% ± 10.83%, n = 15 cells; 4 h, 92.35% ± 9.47%, n = 15 cells; 12 h, 69.49% ± 9.36%, n = 15 cells; 24 h, 145.4% ± 14.32%, n = 15 cells; 48 h, 192.2% ± 23.59%, n = 15 cells. TTX: 0 h, 100% ± 10.66%, n = 15 cells; 4 h, 25.77% ± 3.04%, n = 15 cells; 12 h, 40.22% ± 6.26%, n = 15 cells; 24 h, 26.64% ± 2.31%, n = 15 cells; 48 h, 35.02% ± 4.53%, n = 15 cells) Statistical analysis was performed using a nonparametric one-way ANOVA test. * P < 0.05, ** P < 0.01, and *** P < 0.001 vs 0 h group. Error bars represent s.e.m. (e) Left and right panels, unlabeled WT and labeled Narp^−/− neurons were co-cultured at a ratio of 10:1. Images are representative images of an unlabeled WT (left panel) and labeled Narp^−/− (right panel) neurons from the same population. Middle panel, unlabeled Narp^−/− and labeled WT neurons were co-cultured at a ratio of 10:1. Shown is a representative image of an unlabeled Narp^−/− neuron. Scale bars represent 10 μm (f) Summary of the data shown in e. GluR4 intensity per μm dendrite for all cell types were normalized to WT unlabeled neurons (WT pre/WT post, 100% ± 12.25%, n = 26 cells; ^−/− pre/^−/− post, 35.57% ± 4.66%, n = 25 cells; WT pre/^−/− post, 68.89% ± 8.52%, n = 19 cells; ^−/− pre/WT post, 34.67% ± 10.02%, n = 13 cells). Statistical analysis was performed using a nonparametric one-way ANOVA test. *** P < 0.001 vs. WT pre/WT post. Error bars represent s.e.m.

Figure 6

Narp is required for homeostatic scaling of excitatory synaptic inputs onto PV-INs and regulates their spontaneous firing frequency. (a) Representative mEPSC traces of cultured WT (left) and Narp^−/− (right) Parvalbumin interneurons after 48 hour treatment with TTX (top), vehicle (middle), or bicuculline (bottom). (b) Summary (left) and cumulative probability plot (right) of the mEPSC amplitudes obtained from all recordings similar to those shown in a. (WT TTX, 20.44 pA ± 1.10 pA, n = 17 cells; WT Untreated, 23.45 pA ± 0.85 pA, n = 20 cells; WT Bicuculline, 28.58 pA ± 2.68 pA, n = 15 cells; Narp^−/− TTX, 19.14 pA ± 0.85 pA, n = 13 cells; Narp^−/− Untreated, 19.17 pA ± 0.86 pA, n = 23 cells, Narp^−/− Bicuculline, 20.17 pA ± 1.51 pA, n = 13 cells) Statistical analysis was performed using a Student's t-test. ** P < 0.01, *** P < 0.001 vs. Untreated WT group. Error bars represent s.e.m. (c) Summary (left) and cumulative probability plot (right) of the mEPSC frequency for all PV-INs similar to those shown in a. (WT TTX, 31.06 Hz ± 2.33 Hz, n = 17 cells; WT Untreated, 31.73 Hz ± 2.47 Hz, n = 20 cells; WT Bicuculline, 36.19 Hz ± 2.92 Hz, n = 15 cells; Narp^−/− TTX, 31.73 Hz ± 2.33 Hz, n = 13 cells; Narp^−/− Untreated, 32.66 Hz ± 2.54 Hz, n = 23 cells, Narp^−/− Bicuculline, 33.99 Hz ± 3.62 Hz, n = 13 cells) Statistical analysis was performed using a Student's t-test. Error bars represent s.e.m. (d) Representative current clamp recordings from cultured WT (top) and Narp^−/− (bottom) PV-INs showing the rate of spontaneous action potentials in the absence (left) and presence (right) of 10 μM gabazine, 10 μM NBQX, and 50 μM D-APV. (e) Summary of the spontaneous firing frequency of untreated PV-INs from all recordings similar to those illustrated in d. (WT, 2.67 Hz ± 0.51 Hz, n = 14 cells; Narp^−/−, 1.13 Hz ± 0.18 Hz, n = 24 cells). Statistical analysis was performed using a Student's t-test. ** P < 0.01. Error bars represent s.e.m.

Figure 7

Narp regulates PV-IN synaptic strength in acute hippocampal slices (a) Representative sEPSC records of PV-INs in acute slices from control (left) and MECS administered (right) WT (top) and Narp^−/− (bottom) mice (bars 1 s/100 pA). At right of each trace is also shown the average sEPSC from each record (bars 2 ms/20 pA) (b) Bar chart summary of average sEPSC amplitudes, interevent intervals (IEIs), rise times, and decay time constants obtained from recordings in WT (left) and Narp^−/− (right), unstimulated (black) and MECS administered (red) mice. Also shown is the group data for action potential frequency observed in response to a sustained current injection (0.8 s/800 pA) in current-clamp mode. (WT Control sEPSC amplitude, 26.7 pA ± 2.2 pA, n = 8 cells from 4 mice; WT MECS sEPSC amplitude, 36.7 pA ± 3.2 pA, n = 7 cells from 4 mice; Narp^−/− Control sEPSC amplitude, 27.9 pA ± 3.3 pA, n= 10 cells from 4 mice; Narp^−/− MECS sEPSC amplitude, 31 pA ± 2.2 pA, n = 8 cells from 4 mice). Note the scaling factors (×10 or /10) for several parameters to fit on the same Y axis. Statistical analysis was carried out using a Student's t-test. * P < 0.05. Error bars represent s.e.m. (c) Cumulative probability plot for the amplitudes of all sEPSC events from all recordings obtained in WT and Narp^−/− PV-INs for control and MECS conditions as indicated (for WT, P < 0.01 for control vs MECS). Statistical analysis was carried out using a K-S test.

Figure 8

Narp^−/− mice are hypersensitive to kindling-induced seizures. (a) Narp^−/− mice experienced Class V behavioral seizures after fewer evoked ADs than WT mice, indicating an enhanced rate of kindling progression (WT: 1st CL III/IV AD, 4.5 ± 0.5, 1st CL V AD, 9.3 ± 0.8, 3 CL V ADs, 16.3 ± 1.08, n = 22 mice; Narp^−/−: 1st CL III/IV AD, 4.4 ± 0.5, 1st CL V AD, 5.9 ± 0.9, 3 CL V ADs, 11.0 ± 1.6, n = 7). Statistical analysis was carried out using a one-way ANOVA and a Bonferonni test for multiple comparisons. * P < 0.05 vs. WT. Error bars represent s.e.m. (b) The relative AD threshold (i.e., the stimulation intensity required to evoke the nth AD/the stimulation required to evoke the 1^st AD) decreases more rapidly for the Narp^−/− mice (open squares) relative to WT mice (black diamonds) (WT AD #25, 0.63 ± 0.11, n = 24 mice; Narp^−/− AD #25, 0.37 ± 0.08, n = 9 mice). Error bars represent s.e.m.