Activity-regulated somatostatin expression reduces dendritic spine density and lowers excitatory synaptic transmission via postsynaptic somatostatin receptor 4

Abstract

Neuronal activity regulates multiple aspects of the morphological and functional development of neural circuits. One mechanism by which it achieves this is through regulation of gene expression. In a screen for activity-induced genes, we identified somatostatin (SST), a neuropeptide secreted by the SST subtype of interneurons. Using real time quantitative PCR and ELISA, we showed that persistent elevation of neuronal activity increased both the gene expression and protein secretion of SST over a relatively prolonged time course of 48 h. Using primary hippocampal neuronal cultures, we found that SST treatment for 1 day significantly reduced the density of dendritic spines, the morphological bases of excitatory synapses. Furthermore, the density of pre- and postsynaptic markers of excitatory synapses was significantly lowered following SST treatment, whereas that of inhibitory synapses was not affected. Consistently, SST treatment reduced the frequency of miniature excitatory postsynaptic currents, without affecting inhibition. Finally, lowering the endogenous level of SST receptor subtype 4 in individual hippocampal pyramidal neurons significantly blocked the effect of SST in reducing spine density and excitatory synaptic transmission in a cell autonomous fashion, suggesting that the effect of SST in regulating excitatory synaptic transmission is mainly mediated by SST receptor subtype 4. Together, our results demonstrated that activity-dependent release of SST reduced the density of dendritic spines and the number of excitatory synapses through postsynaptic activation of SST receptor subtype 4 in pyramidal neurons. To our knowledge, this is the first demonstration of the long term effect of SST on neuronal morphology.

FIGURE 1.

Neuronal activity increased the expression and secretion of SST. A, microarray results from three independent hippocampal neuronal culture preparations showed that SST mRNA level was significantly elevated following global activity elevation with bicuculline (Bic.) and blocked by co-application of TTX (Bic., 4 h: 1.43 ± 0.005, p < 0.001; Bic., 4 h + TTX: 1.03 ± 0.02, p > 0.05; Bic., 48 h: 2.14 ± 0.15, p < 0.01; Bic., 48 h + TTX: 0.45 ± 0.06, p < 0.001). B, real time qPCR results from additional culture preparations showing that the SST mRNA level was significantly increased following different manipulations that elevated neuronal activity for 4 or 48 h (Bic., 4 h: 1.66 ± 0.26, p < 0.05; Bic., 48 h: 2.79 ± 0.71, p < 0.05; KA 4 h: 2.05 ± 0.27, p < 0.01; KA 48 h: 3.01 ± 0.41, p < 0.01; KCl 4 h: 2.16 ± 0.31, p < 0.01; KCl 48 h: 5.03 ± 0.73, p < 0.01). C, real time qPCR results showed that SST expression in the hippocampus was up-regulated following KA injection in vivo (KA 6 h: 1.78 ± 0.14, p < 0.01; KA 24 h: 1.72 ± 0.08, p < 0.001). D, ELISA results showed that the level of SST in the medium of hippocampal neuronal cultures was elevated following bicuculline treatment (4 h: 1.02 ± 0.07, p > 0.05; 12 h: 1.36 ± 0. 15, p > 0.05; 24 h: 1.65 ± 0.26, p < 0.05; 48 h: 5.78 ± 1.80, p < 0.05). In all experiments, the mRNA or protein levels were normalized to that of untreated controls. The results are shown as the means ± S.E. n, shown as white numbers in the bars, represents the number of independent culture preparations (A, B, and D) or the number of rats (C). *, p < 0.05; **, p < 0.01; ***, p < 0.001. Ctrl, control.

FIGURE 2.

SST was specifically expressed in interneurons following activity elevation in vitro or in vivo. A, representative images of SST, GABA, and MAP2 immunostaining of DIV 12 primary hippocampal neurons, treatments as indicated. The scale bar is 50 μm. The number of neurons co-labeling with SST and GABA (SST⁺ GABA⁺), as a percentage of total SST immunopositive neurons (SST⁺) is quantitated to the right of each example. B, representative images of the CA1 region of GAD67-GFP mice, untreated or 24 h following KA injection, channels as labeled. The scale bar is 50 μm for the top row of each condition and 20 μm for the zoomed images (white boxes of top images) in the bottom row of each condition. The number of the neurons co-immunostaining with SST and GFP (SST⁺ GFP⁺), as a percentage of total SST immunopositive neurons (SST⁺), is shown to the right of each example. Ctrl, control; Bic., bicuculline.

FIGURE 3.

SST treatment reduced the density of dendritic spine, as well as that of pre- and postsynaptic markers of excitatory synapses, without affecting inhibitory synapses. A, representative images of the dendritic spines of control (Ctrl) neurons or those treated with SST. The scale bar is 5 μm. B, quantitation of dendritic spine density following SST treatment (control: 3.03 ± 0.19; SST: 2.22 ± 0.14, p < 0.01). C, SST treatment did not affect the size of dendritic spines: p > 0.05. D, E, H, and I, representative images of dendrites (labeled with MAP2 or GFP) immunostained with Bassoon and co-stained with vGluT1 (D), surface GluR2 (sGluR2, E), vGAT (H), or GABA_ARα1 (I). The dendritic segment within each white box is shown as a zoomed image below. The scale bar is 20 μm for full frame images and 5 μm for the zoomed images. F, SST treatment reduced the density of synaptic vGluT1 (control: 1 ± 0.05; SST: 0.80 ± 0.07, p < 0.05), without affecting the integrated intensity of vGluT1 puncta (control: 1 ± 0.05; SST: 0.96 ± 0.05, p > 0.05). G, SST treatment reduced the density of synaptic GluR2 (control: 1 ± 0.09; SST: 0.73 ± 0.07, p < 0.05), without affecting the integrated intensity of synaptic GluR2 puncta (control: 1 ± 0.04; SST: 1.02 ± 0.05, p > 0.05). J, SST treatment did not affect the density (control: 1 ± 0.11; SST: 1.04 ± 0.09, p > 0.05) or integrated intensity (control: 1 ± 0.07; SST: 0.94 ± 0.08, p > 0.05) of synaptic vGAT puncta. K, the density (control: 1 ± 0.06; SST: 0.98 ± 0.07, p > 0.05) or integrated intensity (control: 1 ± 0.05; SST: 1.05 ± 0.08, p > 0.05) of synaptic GABA_ARα1 was not affected by SST treatment. The results are shown as the means ± S.E. n, shown as white numbers in the bars, represents the number of neurons. *, p < 0.05; **, p < 0.01.

FIGURE 4.

SST treatment reduced the frequency of mEPSCs, without affecting mIPSCs. A, representative mEPSC traces and average waveforms from Ctrl or SST-treated neurons. B, SST treatment did not affect the amplitude of mEPSCs (control: 15.62 ± 1.05 pA; SST: 14.73 ± 0.78 pA, p > 0.05) but significantly reduced their frequency (control: 1.28 ± 0.18 Hz; SST: 0.64 ± 0.10 Hz, p < 0.01). C and D, cumulative probability plot of mEPSC amplitudes (p > 0.05, C) and interevent intervals (p < 0.01, D). E, representative mIPSC traces and average waveforms from control and SST-treated neurons. F, SST treatment did not affect the amplitude (control: 25.67 ± 1.37 pA; SST: 23.62 ± 0.64 pA, p > 0.05) or frequency (control: 0.75 ± 0.11 Hz; SST: 0.72 ± 0.10 Hz, p > 0.05) of mIPSCs. G and H, cumulative distributions of mIPSC amplitudes (p > 0.05, G) or interevent intervals (p > 0.05, H). The results are shown as the means ± S.E. n, shown as white numbers in the bars, represents the number of neurons. **, p < 0.01.

FIGURE 5.

The effects of SST on spines and mEPSCs were mediated by postsynaptic SSTR4. A, the SSTR4 RNAi (R4 RNAi) was efficient in reducing the level of SSTR4 in hippocampal neurons (R4 RNAi: 0.32 ± 0.08, p < 0.01). B, representative images of dendritic spines, conditions as indicated. The scale bar is 5 μm. C, SSTR4 RNAi blocked SST-induced reduction in dendritic spine density (control: 3.83 ± 0.21; SST: 2.98 ± 0.16, p < 0.05 versus control; R4 RNAi: 3.78 ± 0.22, p > 0.05 versus control; R4 RNAi + SST: 3.99 ± 0.24, p > 0.05 versus R4 RNAi). D, none of the treatments affected the size of dendritic spines: p > 0.05 for all conditions. E, representative mEPSC traces and average waveforms, conditions as indicated. F, SSTR4 RNAi blocked the effect of SST on reducing mEPSC frequency (control: 0.79 ± 0.16 Hz; SST: 0.36 ± 0.05 Hz, p < 0.05 versus control; R4 RNAi: 0.78 ± 0.15 Hz, p > 0.05 versus control; R4 RNAi + SST: 0.79 ± 0.15 Hz, p > 0.05 versus R4 RNAi). G, cumulative distribution of interevent intervals (SST versus control: p < 0.001; R4 RNAi versus control: p > 0.05; R4 RNAi + SST versus R4 RNAi: p > 0.05). H, SSTR4 RNAi did not affect mEPSC amplitude in control or SST-treated conditions. p > 0.05 for all conditions. The results are shown as the means ± S.E. n, shown as white numbers in the bars, represents the number of neurons. *, p < 0.05; **, p < 0.01. Ctrl, control.