Introduction

Activity-dependent gene expression is a fundamental property of neurons by which they modify synapses in response to experience. The mechanisms by which neurons convert activity into gene expression are well characterized. Neuronal activation stimulates Ca2+ entry into the cell, stimulating downstream molecular signaling programs that lead to activation of immediate-early genes such as Fos and Jun (Yap and Greenberg, 2018). However, emerging evidence demonstrates that astrocytes also exhibit activity-dependent gene expression (Hasel et al., 2017; Nagai et al., 2019; Sardar et al., 2023). These transcriptional responses are as robust as many classes of neurons (Hrvatin et al., 2018) and include several genes associated with synapse formation and function (Genoud et al., 2006; Allen and Eroglu, 2017; Farhy-Tselnicker et al., 2021). The precise molecular signaling programs that couple synaptic activity to gene expression in astrocytes are not well defined.

The molecular signaling pathway, Sonic hedgehog (Shh), is emerging as an important mediator of reciprocal interactions between neurons and astrocytes. In the postnatal and adult brain, SHH ligand is produced by neurons which acts directly on neighboring astrocytes, regulating SHH-dependent gene expression programs (Garcia et al., 2010; Farmer et al., 2016; Xie et al., 2022). Notably, genetic ablation of pathway activity, selectively in astrocytes, decreases expression of the glial-specific, inward rectifying K+ channel, Kir4.1, perturbing extracellular K+ homeostasis and increasing neuronal excitability (Hill et al., 2019). Conversely, unrestrained Shh signaling promotes synapse formation mediated by increased expression of synaptogenic cues (Xie et al., 2022). These observations point to SHH as a key mediator of neuron–astrocyte interactions that reciprocally shape synapse number and function. Furthermore, noncanonical Shh signaling in neurons is required for establishing corticocortical synapses (Harwell et al., 2012), suggesting diverse and cell-type–specific actions of SHH on synaptic regulation. Here, we focus on Shh signaling in astrocytes mediated by canonical transduction of the pathway.

Transduction of Shh signaling is mediated by transcriptional activation of its effectors, the GLI family of transcription factors, that regulate expression of SHH target genes (Briscoe and Therond, 2013). The binding of SHH to its receptor, Patched (Ptch1), relieves inhibition of the obligatory coreceptor, Smoothened (Smo), promoting transcription of SHH target genes, including Gli1. High levels of SHH promote transcriptional activation of Gli1, which acts as a reliable readout of pathway activation (Bai et al., 2002). At baseline, the transcriptional abundance of Gli1 in the cortex is low (Xie et al., 2022), and its expression is localized to astrocytes in layers IV and V, consistent with the distribution of SHH-expressing neurons in layer V (Garcia et al., 2010; Harwell et al., 2012). High-frequency stimulation of cultured hippocampal neurons promotes SHH release that is sensitive to TTX treatment (Su et al., 2017), suggesting that neural activity promotes Shh signaling. Interestingly, SHH protein is localized in axons (Traiffort et al., 1998, 1999; Charytoniuk et al., 2002; Peng et al., 2018), where it is associated with synaptic vesicles (Beug et al., 2011) making it well positioned to couple with neuronal activity. Whether physiological neural activity promotes SHH-mediated interactions between neurons and astrocytes in vivo, and how these interactions influence synapses, is not known.

To examine this, we exposed juvenile mice to an enriched somatosensory environment to stimulate active whisking. We show that sensory and chemogenetic stimulation increases Shh signaling in cortical astrocytes. Exposure to an enriched environment (EE) produced an increase in synapse number and a concomitant increase in expression of Hevin and secreted protein acidic and rich in cysteine (SPARC), matricellular proteins produced by astrocytes that modify synapses. Structural plasticity and experience-dependent upregulation of Hevin and SPARC are occluded in astrocyte-specific conditional knock-out (CKO) mice in which Shh signaling is selectively abolished, demonstrating a requirement for Shh signaling in astrocyte modulation of synapses. Taken together, these findings demonstrate novel, activity-dependent regulation of Shh signaling and establish SHH as a molecular mechanism by which astrocytes couple neuronal activity to gene expression and modulate synaptic plasticity.

ResultsNeuronal activity stimulates Sonic hedgehog signaling in vivo

To examine whether neural activity stimulates Shh signaling in vivo, we used a chemogenetic approach to selectively excite cortical pyramidal neurons with the excitatory hM3DGq designer receptor exclusively activated by designer drugs (DREADD; Fig. 1A,B). We injected AAV8-CamKIIα-hM3DGq-mCherry virus (Alexander et al., 2009) into the somatosensory cortex of Gli1nlacZ/+ mice which express nuclear lacZ at the Gli1 locus (Bai et al., 2002). Because Gli1 is a canonical target of Shh signaling, and because SHH is required for Gli1 expression, transcriptional activation of Gli1 acts as a reliable readout of Shh signaling (Bai et al., 2002; Hui and Angers, 2011). After 14 d, we treated mice with medicated drinking water containing the designer drug clozapine N-oxide (CNO) for 7 d to chronically activate the receptor (Armbruster et al., 2007). Transfected cells expressed c-Fos (Fig. 1C), and CNO-treated mice showed a significant increase in the number of c-Fos-labeled cells compared with virus-injected animals that were fed unmedicated drinking water, indicating successful activation of neurons (Fig. 1D,E). We observed a significant increase in the number of β-Gal+ cells in the somatosensory cortex in the CNO-treated group compared with untreated controls (Fig. 1F,G). Ninety-seven percent of β-Gal cells colocalized with Sox9 (Fig. 1H,I), a well-established marker for identifying astrocytes (Sun et al., 2017), consistent with our previous work identifying astrocytes as the predominant cell type expressing Gli1 in the mature forebrain (Garcia et al., 2010; Gingrich et al., 2022). The fraction of cells identified as astrocytes does not change between water and CNO groups (Fig. 1H,I). Taken together, these data demonstrate that neuronal excitation stimulates Shh signaling in astrocytes in vivo.

Figure 1.

Neuronal activity stimulates Shh signaling. A, Schematic depicting experimental approach. Gli1nlacZ/+ mice were injected with AAV8-CamKIIα-hM3DGq-mCherry in the somatosensory cortex. Two weeks later, mice were fed CNO-medicated or unmedicated drinking water for 7 d before analysis. B, The transfected region can be visualized with mCherry (magenta). Counterstained with DAPI (blue). SSCtx, somatosensory cortex. Scale bar, 250 μm. C, mCherry (magenta, left panel) and c-Fos (green, middle panel) in the transfected region. Arrows point to cells coexpressing mCherry and c-Fos. Scale bar, 25 μm. D, F, Bright-field immunohistochemistry for c-Fos (D) and β-Gal (F) in the somatosensory cortex. Scale bar, 25 μm. E, G, Stereological quantification of c-Fos (E) and β-Gal (G) in the transfected region shows increased neural activity and increased Shh signaling in mice that received CNO compared with water-fed controls. n = 5 mice in water condition, n = 4 mice in CNO condition. Statistics: Student's t tests. H, Immunofluorescent staining for Sox9 (magenta) and β-Gal (green) in the transfected region between water (left) and CNO (right) groups. Arrows point to β-Gal cells colabeled with Sox9. Scale bar, 25 μm. I, Fraction of β-Gal-labeled cells colocalizing with Sox9 comparing between water and CNO groups. n = 3 mice in water condition, n = 3 mice in CNO condition. Statistics: Student's t test. In all graphs, data points represent individual animals; bars show mean ± SEM. Extended Data Figure 1-1 shows that chemogenetic stimulation does not increase the number of SHH-expressing neurons.

Figure 1-1

Chemogenetic stimulation of neuronal activity does not increase the number of Shh-expressing neurons. (A) Schematic depicting experimental approach to determine whether chemogenetic stimulation of neuronal activity increases the number of Shh-expressing cells. ShhCreER/+; Ai14 mice were injected with AAV5-hSyn-hM3DGq-HA targeting the somatosensory cortex. After 2 weeks, mice were fed CNO-medicated or unmedicated water for 10 days, receiving 3 doses of tamoxifen during the last 3 days. Tissues were analyzed seven days later. dpi, days post injection. (B-D) Immunolabeling for HA (green) shows the transduced region overlapping with tdTom labeled (magenta) cells in the cortex. Inset shown in (C). Individual cell shown at high power in (D). Counterstained with DAPI (blue). Scale bar, 25 μm. (E) The number of Shh neurons in the cortex between water control versus CNO. Data points represent individual animals, n = 2 mice in water group, n = 2 mice in CNO group. Bars show mean ± SEM. Statistic: Student’s t-test. Download Figure 1-1, TIF file.

In the cortex, Shh is expressed predominantly in a subset of excitatory pyramidal neurons in layer V (Harwell et al., 2012; Hill et al., 2019). To determine whether stimulation of neuronal activity increases the number or distribution of Shh-expressing cells, we also performed chemogenetic activation of cortical neurons in ShhCreER/+;Ai14 mice. We injected mice with AAV5-Syn-hM3DGq-HA virus and began CNO treatment 2 weeks later. Animals received CNO or unmedicated water for 10 d. To mark Shh-expressing cells, animals received tamoxifen over the last 3 d of CNO treatment and were analyzed 2 weeks later (Extended Data Fig. 1-1). There was no difference in the number or distribution of Shh-expressing cells between the CNO-treated and control animals (Extended Data Fig. 1-1), indicating an increase in the availability of SHH ligand, rather than an increase in the number of Shh-expressing neurons.

Sensory experience stimulates Sonic hedgehog signaling

We next examined whether a more physiological stimulus can increase Shh signaling. To test this, we placed Gli1nlacZ/+ mice in an enriched sensory environment designed to stimulate active whisking (Yang et al., 2009). Mice were weaned into large rat cages from which strings of beads were hung from the wire cage lid in a manner requiring constant interaction at Postnatal day 21 (P21). Control mice were weaned into similar rat cages with only standard nesting materials and no beads. Mice were housed in the enriched environment (EE) or standard housing (SH) conditions for 3 weeks (Fig. 2A). A significant increase in the number of cells labeled with c-Fos was observed in the somatosensory cortex of mice housed in EE compared with SH conditions (Fig. 2C), confirming increased activity in this region. Mice housed in EE showed a significant increase in the number of β-Gal-labeled cells compared with control mice housed in SH conditions (Fig. 2B,D), indicating an increase in Shh activity in response to sensory experience. Notably, there was no difference in either c-Fos or β-Gal labeling between EE and SH conditions in the visual cortex (Fig. 2E,F), demonstrating that sensory stimulation does not produce a generalized increase in Shh signaling but rather selectively increases activity of the pathway in circuits that are stimulated by experience. Because sensory activity stimulates Shh signaling, we next asked whether sensory deprivation lowers the activity of the pathway. To do this, we performed unilateral whisker trimming on P21 Gli1nlacZ/+ mice for 3 weeks, starting at birth. We analyzed the deprived, contralateral hemisphere and compared it to the intact, ipsilateral hemisphere. We observed significantly fewer β-Gal+ cells in the contralateral, deprived hemisphere compared with the ipsilateral, intact hemisphere (Fig. 2G). Analysis of the auditory cortex showed whisker trimming had no effect on Shh signaling in this region (Fig. 2H). Taken together, these data demonstrate that activation of distinct sensory circuits selectively stimulates Shh signaling in cortical astrocytes in a circuit-specific manner.

Figure 2.

Enriched experience stimulates Shh signaling. A, Experimental timeline. Gli1nlacZ/+ mice were weaned into standard housing (SH) or enriched experience (EE) cages at P21 and analyzed at P42. B, Representative bright-field immunohistochemistry for β-Gal in the somatosensory cortex. Scale bar, 50 μm. C–F, Stereological quantification of cells expressing the immediate-early gene c-Fos (C, E) or immunolabeled with β-Gal (D, F) in the somatosensory (C, D) or visual (E, F) cortices. n = 4–10 mice in SH, n = 6–10 mice in EE. Statistics: Student's t tests. G, H, Gli1nlacZ/+ mice were subjected to unilateral whisker trimming continuously from P21 and analyzed at P42. Stereological quantification of β-Gal cells from the ipsilateral (intact) and contralateral (deprived) barrel (G) and auditory (H) cortex. n = 3 mice. Statistics: paired t test. I, Schematic depicting genetic strategy to label cells with temporally distinct Shh signaling (left panel) and timeline of experiment (right panel). In Gli1nlacZ/+ mice, β-Gal expression labels cells with active Shh signaling. In Gli1CreER;Ai14 mice, Cre-mediated recombination promotes permanent expression of tdTom. In Gli1CreER/nlacZ;Ai14 mice, tdTom expression identifies cells with Shh signaling at baseline (t0) and β-Gal expression identifies Shh signaling in cells at the end of the experiment (t1). Double-labeled cells identify cells with persistent Shh signaling throughout the experiment. J, K, Low power (J) and high power (K) images showing tdTom expression (magenta, left panel) and immunostaining for β-Gal (green, middle panel; merged image, right panel) in the cortex of mice housed in EE as described in I. Scale bar, 250 μm. K, Individual cells show those with upregulation (blue arrowhead), downregulation (open arrowhead), and persistent (white arrow) Shh activity. Scale bar, 50 μm. L–N, The fraction of β-Gal-labeled cells that are single labeled (upregulated Shh activity; L), tdTom-labeled cells that are single labeled (downregulated Shh activity; M), and all cells that are double labeled (persistent Shh activity; N) in SH or EE. n = 4 mice in SH, n = 3 mice in EE. O, Schematic depicting experimental approach: P21 ShhCreER/+;Ai14 mice were weaned into SH or EE cages and received three doses of tamoxifen from P39 to P41. Tissues were analyzed 2 weeks after the last tamoxifen dose. P–R, Low power image showing tdTom-labeled cells (magenta) in ShhCreER/+;Ai14 mice. Labeled cells are found predominantly in layer V. Inset shows individual cells with neuronal morphology in the cortex (Q). Note the absence of labeled cells in the VPM (R). Counterstained with DAPI (blue). Scale bars: P, 50 μm; Q, R, 20 μm. S, The number of Shh neurons in the cortex in SH and EE mice. n = 3 mice in SH, n = 3 mice in EE. In all figures, data points represent individual animals; bars show mean ± SEM.

To precisely identify cells that show increased Shh activity following experience, we combined the direct reporter approach with a tamoxifen-dependent labeling strategy, enabling us to monitor temporally distinct populations of cells experiencing Shh signaling. We crossed Gli1nlacZ/+ mice with Gli1CreER/+;Ai14 mice carrying the Cre-dependent Ai14 reporter allele expressing tdTomato (tdTom; Gli1nlacZ/CreER;Ai14), enabling identification of cells with Shh signaling at the start and conclusion of sensory enrichment. Because lacZ expression is directly regulated by the transcriptional activity of Gli1, cells actively transducing SHH can be monitored by immunohistochemistry for β-Gal, the protein product of the lacZ gene. In contrast, tamoxifen-dependent, Cre-mediated recombination produces permanent expression of tdTom in cells, even after Shh activity is downregulated (Fig. 2I). Thus, in Gli1nlacZ/CreER;Ai14 mice, lacZ expression reflects active Shh signaling whereas tamoxifen-dependent Cre-mediated expression of tdTom reflects historical signaling. Although these mice are effectively Gli1 nulls, GLI1 is dispensable for Shh signaling due to the redundant activator function of GLI2 (Bai et al., 2002; Niewiadomski et al., 2019), and we did not observe any gross anatomical or behavioral phenotypes, consistent with previous studies (Bai et al., 2002; Gingrich et al., 2022). To mark cells with Shh signaling at baseline, mice received tamoxifen before placing them in EE or SH housing for 3 weeks (Fig. 2I). We first examined the population of β-Gal+ cells and found a large fraction of cells that were negative for tdTom (Fig. 2J,K), indicating Shh activity in a population of cells distinct from those marked at the start of the experiment. Notably, there was a significant increase in single-labeled β-Gal cells after EE compared with SH (Fig. 2L). Whereas 59% of β-Gal-labeled cells were single labeled in SH mice, this fraction increased to 76% in EE mice (Fig. 2L) demonstrating experience-dependent activation of the pathway in cells distinct from those at baseline. These cells were found in layers IV and V (Fig. 2J), suggesting that experience-dependent Shh activity remains localized to deep cortical layers, consistent with the localization of Shh-expressing neurons in layer V (Garcia et al., 2010; Harwell et al., 2012). We next examined the population of cells expressing tdTom. tdTom+ cells were localized primarily in layers IV and V and showed a bushy morphology characteristic of protoplasmic astrocytes, consistent with previous studies (Garcia et al., 2010; Hill et al., 2019; Fig. 2J,K). In SH mice, a large proportion of tdTom+ cells did not colabel with β-Gal (Fig. 2K,M), indicating downregulation of pathway activity. This fraction did not change after exposure to EE (Fig. 2M). We also observed a third population of cells, those coexpressing both tdTom and β-Gal, suggesting either persistent or recurrent Shh activity in these cells (Fig. 2K,N). This fraction was similar between SH and EE conditions (Fig. 2N). These data suggest that Shh activity in individual cells is highly dynamic and that over a 3-week period, individual cells are not likely to experience persistent Shh signaling.

To examine whether enriched experience stimulates Shh expression, we housed ShhCreER/+;Ai14 mice in EE or SH conditions. Mice received tamoxifen over the last 3 d of the housing experience and were analyzed 2 weeks later (Fig. 2O). Similar to our chemogenetic approach, labeled neurons were found primarily in layer V (Fig. 2P,Q), and we found no difference in the number of Shh-expressing neurons between SH and EE (Fig. 2S). These data demonstrate that while transduction of Shh signaling in astrocytes is dynamically modulated between individual cells, the SHH ligand is derived from a defined population of neurons. The increase in Shh signaling may arise from an increase in SHH release or from increased diffusion in the extracellular space. Because astrocytes in layer IV show high levels of Shh activity, we also examined the ventral posterior medial nucleus of the thalamus (VPM) where thalamocortical neurons projecting to layer IV of the barrel cortex reside (Staiger and Petersen, 2020). We detected few, if any, labeled neurons in this region (Fig. 2R) suggesting that Shh signaling in cortical astrocytes is initiated by ligand released from local neurons.

We examined the identity of β-Gal-labeled cells with cell-type–specific markers. Colocalization with Sox9, CAII, and NeuN to identify astrocytes, oligodendrocytes, and neurons, respectively, showed that 96% of β-Gal-labeled cells correspond to astrocytes in both SH and EE (Fig. 3A–C,E) consistent with our previous observation that astrocytes are the primary targets of canonical Shh signaling in the adult brain (Garcia et al., 2010). Β-Gal+ cells did not express Ki-67 (Fig. 3D), ruling out the possibility that the observed increase in cell number is due to proliferation.

Figure 3.

Enriched experience stimulates Shh activity in astrocytes. A–D, Immunofluorescence for β-Gal (green) and Sox9 (A), NeuN (B), CAII (C), or Ki-67 (D; magenta) in the cortex of Gli1nlacZ/+ mice. Counterstained with DAPI (blue). Merged images shown in the right panels. Scale bar, 25 μm. E, The fraction of β-Gal-labeled cells colocalizing with Sox9 does not change between SH and EE. Data points represent individual animals, n = 4 mice in SH, n = 4 mice in EE, 100–400 β-Gal cells analyzed per animal. Bars show mean ± SEM. Statistic: Student's t test.

Sensory experience dynamically modulates Sonic hedgehog signaling

We next examined whether the experience-induced increase in Shh activity reflects a long-term change in Shh signaling. We first determined whether a much shorter time frame was sufficient to increase Shh activity. We housed Gli1nlacZ/+ mice in either SH or EE for 2 d starting at P21. Mice housed in EE showed a significant increase in β-Gal+ cells compared with SH controls (Fig. 4A,B), suggesting that brief exposure to sensory enrichment is sufficient to stimulate Shh signaling. To examine whether experience-dependent Shh signaling persists after removal of the stimulus, we housed Gli1nlacZ/+ mice in EE for 2 d, then housed them in SH condition for 2 d, and compared Gli1 activity with mice continuously housed in EE or SH conditions for 4 d. To rule out the possibility that any observed differences in β-Gal number may reflect perdurance of the protein rather than sustained Shh activity, we used single-molecule fluorescent in situ hybridization (smFISH) by RNAscope to more directly measure Gli1 transcripts (Fig. 4C). In mice continuously housed in EE, we found an increase in cells expressing Gli1 compared with SH control mice (Fig. 4C,D). However, in mice returned to SH after 2 d of EE, Gli1 expression returned to baseline (Fig. 4C,D). Notably, individual cells expressing Gli1 also showed a strong increase in the number of transcripts in the nucleus after exposure to EE compared with SH-housed mice (Fig. 4C,E,F). This suggests that neural activity not only stimulates Shh signaling in new cells but also further increases the activity of the pathway in cells with baseline Shh signaling. These data demonstrate that Shh signaling is highly dynamic and responsive to changing experiences.

Figure 4.

Shh signaling responds rapidly to sensory activation and is not persistent. A, Bright-field immunohistochemistry for β-Gal in the somatosensory cortex of Gli1nlacZ/+ mice housed in SH (top panel) or EE (bottom panel) for 2 d. Scale bar, 50 μm. B, Stereological quantification of β-Gal shows increased Shh activity in the somatosensory cortex after 2 d of EE. Bars represent mean ± SEM. Data points represent individual animals, n = 4 mice in SH, n = 6 mice in EE. Statistics: Student's t test. C, Representative images of fluorescent in situ hybridization for Gli1 mRNA in the deep layers of the cortex from SH, EE, and EE and SH housing conditions. Dotted line outlines individual DAPI nuclei with Gli1 mRNA. Scale bar, 10 μm. D, Gli1+ cell count in layers IV and V of the cortex from mice housed in SH or EE for 4 d or EE for 2 d followed by SH for 2 d (EE + SH). In each condition, 1,000–2,000 DAPI cells were analyzed. Bars represent mean ± SEM. Data points represent individual animals, n = 3 mice per condition. Statistics: one-way ANOVA with Tukey's multiple comparisons. E, F, Quantification of Gli1 puncta in individual cells in layers IV and V from SH, EE, and EE and SH conditions. Bar graphs (E) show data by animal, and violin plots (F) show data by individual cells. In each condition, 400–600 cells were analyzed, and only cells with >3 puncta were included in quantification; n = 3 mice per condition. Bars (E) represent mean ± SEM, and violin plots (F) show median ± interquartile ranges. Statistics: one-way ANOVA with Tukey's multiple comparisons.

To determine whether experience-dependent Shh signaling promotes morphological plasticity of astrocytes, we analyzed the morphologies of tdTom-labeled astrocytes in the barrel cortex of P21 Gli1CreER;Ai14 mice after SH or EE. We reconstructed astrocytes in 3D using Imaris and found no differences in total volume, branch level, process length, or morphological complexity, as measured by Sholl analysis (Fig. 5) between SH and EE conditions.

Figure 5.

Enriched experience does not alter Gli1 astrocyte morphology. A, Representative images for Gli1 astrocytes for mice housed in SH (top) or EE (bottom) in the somatosensory cortex. B–D, Morphological quantification of volume (B), process length (C), and branch level (D). Violin plots show data by individual astrocytes; median ± interquartile, n = 3 mice in SH, n = 3 mice in EE, five astrocytes analyzed per animal. Statistics: Student's t tests. E, Sholl analysis of astrocytes from SH versus EE conditions. Data represent mean ± SEM by animal; n = 3 animals, five astrocytes analyzed per animal.

Astrocytic Sonic hedgehog signaling is required for deep-layer synaptic plasticity

Sensory experience promotes structural plasticity of cortical synapses (Yang et al., 2009). To determine whether experience-dependent Shh signaling in astrocytes promotes synaptic plasticity, we examined dendritic spines of layer V cortical neurons. Excitatory synapses are localized predominantly on dendritic spines (Nimchinsky et al., 2002), which demonstrate remarkable plasticity in response to sensory experience (Zuo et al., 2005; Yang et al., 2009; Landers et al., 2011; Jung and Herms, 2014; Chen et al., 2015). To determine whether Shh signaling plays a role in astrocyte modulation of experience-dependent plasticity, we examined the spine density of cortical neurons in conditional knock-out (CKO) mice lacking Smo, the obligatory coreceptor for transduction of Shh signaling, selectively in astrocytes. We used Gfap-Cre;Smofl/fl mice (Gfap Smo CKO) in which Smo is deleted in Gfap-expressing cells, effectively abolishing Shh signaling in all astrocytes beginning at birth (Garcia et al., 2010; Hill et al., 2019). These mice also carry a Thy1-GFPm allele which labels a subset of layer V pyramidal neurons (Feng et al., 2000), enabling the visualization of individual dendrites and spines. Gli1 expression measured by qRT-PCR showed a 90% reduction in Gfap Smo CKO mice compared with WT littermate controls lacking Cre, confirming effective ablation of Shh activity (Fig. 6A). We analyzed apical dendrites of layer V pyramidal neurons in the barrel cortex, focusing on dendritic segments in layers IV and V where high levels of Shh activity are found. WT (Cre-negative littermates) mice housed in EE conditions showed a significant increase in the density of protrusions compared with control SH mice (Fig. 6B,D), demonstrating that enriched sensory experience promotes structural synaptic plasticity, consistent with previous studies (Yang et al., 2009; Landers et al., 2011). We previously demonstrated that Gfap Smo CKO mice showed an elevated spine density arising from deficits in the developmental elimination of synapses (Hill et al., 2019). Here, we find that sensory enrichment did not produce any further increases in spine density (Fig. 6C,D), consistent with a requirement for astrocytic Shh signaling in structural plasticity. Alternatively, this may instead reflect a limit beyond which no further spines can be added. Shh signaling acts on astrocyte progenitor cells in the perinatal subventricular zone that generate cortical astrocytes (Gingrich et al., 2022). Because recombination begins at P0 in cortical astrocyte progenitors in this Gfap-Cre line (Garcia et al., 2010), this impaired plasticity may reflect developmental perturbation of cortical astrocyte development.