Peroxisome Proliferator‑Activated Receptor Gamma Coactivator‑1α Deficiency in Hippocampal Astrocytes Underlies Enhanced Fear Memory Retrieval in Male Posttraumatic Stress Disorder Model Mice

TL;DR

Reduced PGC-1α in astrocytes of PTSD mice impairs brain communication, leading to stronger fear memories, which can be reversed with treatments targeting these cells.

Contribution

Identifies astrocytic PGC-1α/CX43 as a novel regulatory axis in PTSD-related fear memory retrieval.

Findings

PGC-1α deficiency in hippocampal astrocytes correlates with enhanced fear memory retrieval in PTSD model mice.

Chemogenetic activation of astrocytes or PGC-1α agonists rescues CX43 expression and normalizes fear behavior.

Pharmacological inhibition of CX43 gap junctions mimics the pro-fear effects of PGC-1α deficiency.

Abstract

Peroxisome proliferator‑activated receptor gamma coactivator‑1α (PGC‑1α), a key metabolic regulator, is implicated in astrocyte function, but its specific role during fear memory retrieval and in posttraumatic stress disorder (PTSD) pathogenesis remains unclear. Here, using the single‐prolonged stress mice model, we demonstrated a significant downregulation of PGC‑1α specifically within hippocampal astrocytes, concomitant with reduced astrocyte density, attenuated intracellular Ca2
+ signaling, and impaired activity‑dependent ATP release. Targeted knockdown of hippocampal astrocytic PGC‑1α in vivo was sufficient to potentiate fear memory retrieval. This behavioral enhancement was associated with a loss of complex astrocyte morphology, further suppression of ATP release, aberrant hippocampal neuronal activity, and a marked decrease in connexin 43 (CX43) expression. Notably, pharmacological inhibition of CX43‑based gap junctions mimicked this pro‑fear phenotype in control animals. Conversely, interventions either via chemogenetic activation of astrocytes or administration of PGC‑1α agonists effectively normalized fear memory responses and rescued CX43 levels in PTSD mice. Crucially, the therapeutic benefit of chemogenetic astrocyte activation was abolished when PGC‑1α was concomitantly knocked down. Collectively, our results demonstrate that PGC‐1α deficiency in hippocampal astrocytes underlies enhanced fear memory retrieval in PTSD model mice, highlighting the potential of astrocyte‑targeted strategies for preventing and treating PTSD.

PGC‑1α downregulation in hippocampal astrocytes disrupts ATP release and reduces connexin 43 (CX43) expression, impairing neuro‑glial communication. This promotes enhanced fear memory retrieval in PTSD. Chemogenetic activation of astrocytes or PGC‑1α agonist treatment rescues both CX43 expression and fear behavior, identifying astrocytic PGC‑1α/CX43 as a key regulatory axis and potential target for PTSD intervention.

Figures

The involvement of hippocampal ASTs in SPS‐induced PTSD‐like behaviors. (A) Analysis of the functional connectivity network characteristics of the hippocampus; (B) scanning each sampling volume (corresponding to four slices) in the height direction of 1.6 mm within 2.2 s; (C) brain regions with mean values from both groups and where there was a difference in functional connectivity (filtering condition: difference p < 0.05, correlation ecoefficiency > 0.15) (n = 3/group); (D) regional functional intensity (absolute difference between the average values of control group and model group mice); the involved brain regions include dentate gyrus (DG), thalamus (Tha), olfactory areas (Olf), perirhinal area (Per), striatum dorsal region (DS), visceral area (Vis), field CA1 (CA1), striatum‐like amygdalar nuclei (Str‐amy), hippocampal region (Hip), primary somatosensory area (SI), hypothalamic medial zone (Hyp), retrohippocampal region (Retro‐Hip), pallidum, dorsal region (Dor), primary somatosensory area (Pos), retrosplenial area (Rsp), perirhinal area (Per), hypothalamic lateral zone (Hyp), supplemental somatosensory area (SMA), agranular insular area (Agr), posterior parietal association area (Pos), temporal association areas (Tem), auditory areas (Aud); (E) diagram showing the number of astrocytes in local brain regions of mice; (F) statistical analysis of the differences in the number of astrocytes in various brain regions; (G) representative images of S100A10 immunofluorescence staining in the CA1 regions of the control group and SPS group (scale bar: 50 µm); (H) percentage of colabeling of S100A10 and GFAP in the control group and SPS group (n = 3/group); (I) statistical values of the average fluorescence intensity of S100A10 in the control group and SPS group. Data are presented as mean and SEM. *p < 0.05 between groups (t‐test).

Hippocampal AST calcium signals and ATP release were reduced during fear memory retrieval in PTSD model mice. (A) Flowchart of the experiment (n = 8/group); (B) representative graph of GcaMP6f‐labeled ASTs; (C) representative graph of the calcium signal changes in the hippocampus of mice during the contextual fear test; (D) difference graph of the calcium transient AUC of ASTs; (E) quantitative data of the calcium transient AUC of AST; (F) peak number of calcium signal releases in the contextual fear test of mice; (G) freezing time of mice in the contextual fear test; (H) correlation analysis between the peak number of calcium signal and the freezing time of mice; (I) flowchart of the experiment (n = 6/group) and representative graph of ATP fluorescent probe virus‐labeled AST; (J) representative graph of the ATP fluorescence signal changes in the hippocampus of mice during the contextual fear test; (K) difference graph of ATP fluorescence signal AUC; (L) quantitative data of ATP fluorescence signal AUC; (M) peak mean of ATP fluorescence signal release; (N) peak number of ATP fluorescence signal releases; (O) freezing time of mice in the contextual fear test. Data are presented as mean and SEM. *p < 0.05, p < 0.01, *p < 0.001 between groups (t‐test).

Reduction of AST PGC‐1α–CX43 axis in SPS mice was responsible for the enhanced fear retrieval. (A) Experimental steps of FACS (n = 3/group); (B) gate strategy and sorting results; (C) representative bands and quantitative data of PGC‐1α and NRF1 proteins; (D) representative bands and quantitative data of hippocampal CX43 protein in the control group and on Days 1, 7, and 14 after SPS modeling; (E) flowchart of the experiment (n = 6/group); (F and G) freezing time of Con+CSF and Con+GAP27 group mice in the contextual fear test (F) and the cued fear test (G); (H) ELISA for detecting the ATP levels in the hippocampal tissues of the mice in the Con+CSF and Con+GAP27 groups; (I) ELISA for detecting the ATP levels in the hippocampal tissues of the mice in the control and SPS groups. Data are presented as mean and SEM. *p < 0.05, *p < 0.001 between groups (t‐test or Tukey test).

AST PGC‐1α knockdown enhanced fear memory and anxiety‐like behaviors and impaired AST structure. (A) Flowchart of the experiment (n = 9/group); (B) representative images of virus‐labeled astrocytes; (C) representative bands and quantitative data of PGC‐1α and CX43; (D) freezing of mice in the contextual fear test; (E) freezing of mice in the cued fear test; (F) the percentage of time that mice spent exploring the open arm in the elevated maze test out of the total time; (G and H) the total distance and representative traces of the open field test; (I and J) the results and representative images of the Sholl analysis of astrocytes. The data are presented as mean ± standard error (n = 3/group). (K) Representative image of S100A10 immunofluorescence staining (scale bar: 50 µm); (L) the percentage of colabeling of S100A10 and PGC‐1α (n = 3/group). Data are presented as mean and SEM. *p < 0.05, p < 0.01, *p < 0.001 between groups (t‐test).

AST PGC‐1α knockdown impaired hippocampal neuronal activity. (A) Flowchart of the experiment; (B) representative images showing the colocalization of the fluorescence of the PGC‐1α knockdown virus (green) with the immunofluorescence of the AST marker GFAP (red) and the electrode implantation location (scale bar: 50 µm); (C–E) mean frequency (C), bursts per minute (D), and AI value (E) of mouse hippocampal neurons recorded by multichannel electrodes during the conditioned fear test (n = 5/group); (F) grating diagram of mouse hippocampal neurons spike; (G) schematic of the patch‐clamp experiment (n = 5/group); (H) original current representation of action potential; (I) statistical data of action potential; (J) original current representation of spontaneous excitatory postsynaptic current; (K) frequency of spontaneous excitatory postsynaptic current; (L) amplitude of spontaneous excitatory postsynaptic current. Data are presented as mean and SEM. *p < 0.05, p < 0.01 between groups (t‐test).

AST PGC‐1α knockdown prohibited the release of extracellular ATP. (A) Flowchart of the experiment (n = 4/group); (B) representative images of virus‐labeled astrocytes (scale bar: 50 µm); (C) representative graphs and quantitative data of the area under curve (AUC) of ATP fluorescence signal during the first training period; (D and E) representative graphs showing the changes in ATP levels in the two groups of mice during the contextual fear memory (D) and cued fear memory (E) tests; (F) representative graphs and heat maps of the AUC of ATP levels during the contextual fear memory test; (G) quantitative data of the AUC of ATP levels during the contextual fear memory test; (H and I) peak values (H) and frequencies (I) of ATP release during the contextual fear memory test; (J) representative graph and heatmap of the AUC of ATP levels in the two groups of mice during the fear memory test; (K) quantitative data of the AUC of ATP levels during the fear memory test; (L and M) peak values (L) and frequency (M) of ATP release during the fear memory test. Data are presented as mean and SEM. *p < 0.05 between groups (t‐test).

Activation of hippocampal AST or PGC‐1α reduced fear memory retrieval in SPS mice. (A) Flowchart of experiment 5 (n = 7/group); (B) representative image of immunofluorescence of hM3Dq‐labeled AST; (C) freezing time of mice in the contextual fear test; (D) freezing time of mice in the cued fear test; (E) percentage of time that mice spent exploring the open arm in the elevated maze test out of the total time; (F) flowchart of the experiment 5 (n = 6/group); (G) freezing time of mice in the contextual fear memory; (H) freezing time of mice in the cued fear memory; (I) percentage of time that mice spent exploring the open arm in the elevated maze test out of the total time; (J and K) representative bands of GFAP (J) and CX43 (K) along with quantitative data; (L) flowchart of the experiment 6 (n = 7/group); (M) schematic diagram of the hM3Dq–mCherry/AAV5–PGC‐1α viral injection site and colabeling; (N) freezing time of mice in the contextual fear memory; (O) freezing time of mice in the cued fear memory. Data are presented as mean and SEM. *p < 0.05, p < 0.01, *p < 0.001 between groups (t‐test or Tukey test).

Full text

Introduction

1

The core pathological feature of posttraumatic stress disorder (PTSD) is the abnormal retrieval of fear memory triggered by context‐similar cues, manifesting as pathological generalization of fear memory and impaired extinction ability [1, 2]. Furthermore, neurofunctional imaging has demonstrated that PTSD is characterized by a reduction in hippocampal volume and attenuated activation during fear stimulation [3]. Notably, recent report also indicates that there is a pyramidal neuron–interneuron microcircuit in the ventral hippocampal CA1 region of mice that regulates fear behavior, and correcting this circuit can improve fear behavior [4]. However, systematic study on the regulation of fear memory retrieval by the hippocampus is still relatively limited, especially the exploration of how the hippocampus regulates fear memory retrieval through complex mechanisms such as neuron–glial cell synergy still lags behind that of the amygdala. This research imbalance may limit the complete understanding of the core pathological link of fear memory retrieval and generalization in PTSD.

Astrocytes (ASTs), the predominant glial cell population in the central nervous system, play a pivotal role in sustaining neuronal functional stability and modulating neuronal activity [5, 6]. In PTSD research, ASTs are recognized as important target cells for both pathogenesis and intervention [7, 8]. Optogenetic activation of hippocampal ASTs has been shown to reduce fear memory retrieval in mice [9]. Our experimental evidence also revealed that in PTSD model mice, the number and length of AST protrusions in the hippocampus were reduced compared with control mice [10]. Given these findings, further investigation into the activity and functional changes of hippocampal ASTs during fear memory retrieval, as well as the underlying mechanisms involved in PTSD, is warranted.

Peroxisome proliferator‐activated receptor gamma coactivator‐1α (PGC‐1α) is a key nuclear receptor transcriptional coactivator and a multifunctional regulator of cellular energy metabolism, primarily governing mitochondrial biogenesis, energy metabolism, and antioxidant defense mechanisms [11, 12, 13]. In recent years, its role in the nervous system, particularly in cognitive function regulation, has garnered increasing attention. In neurons, PGC‐1α contributes to the pathogenesis of neurodegenerative diseases such as Alzheimer's disease, vascular dementia, and Parkinson's disease by modulating mitochondrial function, providing antioxidant defense, regulating neuroinflammation, and promoting neural plasticity [14]. In microglia, PGC‐1α plays a crucial role in protecting against brain injury caused by ischemia by alleviating neuroinflammation through regulating autophagy and mitophagy and inhibiting the activation of NOD‐like receptor thermal protein domain associated protein 3 [15]. Recent study has also found that PGC‐1α is highly expressed in ASTs, and PGC‐1α in ASTs plays an important role in regulating AST maturation and synapse formation [16]. Conditional knockout of PGC‐1α in developing ASTs affects AST morphogenesis and synapse formation [16]. Notably, a study on the transcriptome and proteome changes in brain tissues of clinical PTSD patients revealed significant downregulation of cerebral PGC‐1α [17]. Similarly, our team also observed a downward trend in hippocampal PGC‐1α expression in PTSD model mice using the single‐prolonged stress (SPS) mice model [12]. However, research on PGC‐1α in hippocampal ASTs remains limited, and it remains unclear whether and how PGC‐1α in ASTs is involved in fear memory retrieval and PTSD development.

In this study, brain functional ultrasound (fUS) imaging and full‐field scanning were used to analyze changes of ASTs in the hippocampal region of a PTSD mouse model induced by the SPS protocol [18, 19]. Focusing on the hippocampus, we examined whether ASTs were involved in the retrieval of PTSD‐related fear memory. We then designed a shRNA virus specifically targeting the transcriptional region of PGC‐1α in ASTs and assessed the effects of PGC‐1α knockdown (KD) on fear memory, hippocampal AST morphology, and synaptic function. Furthermore, we validated the effects of chemogenetic activation of hippocampal ASTs and PGC‐1α agonist treatment on fear memory in PTSD mice. Our findings demonstrate that PGC‐1α serves as a critical target in PTSD, likely influencing fear memory retrieval through the modulation of AST activity, thereby offering valuable insights for the development of preventive and therapeutic strategies for PTSD.

Results

2

The Involvement of Hippocampal AST in SPS‐Induced PTSD‐Like Behaviors

2.1

Initially, we assessed the connectivity between different brain regions by evaluating the correlation of spontaneous hemodynamic fluctuations. The results showed that the hippocampus had dense functional connections with key brain regions such as the cortex, thalamus, hypothalamus, and amygdala (Figure 1A,B). In SPS mice, the functional connectivity strengths between the hippocampal CA1, CA2, and DG regions and the thalamus and hypothalamus were significantly lower than those in the control group (p < 0.05, F (1, 88) = 12.69; Figure 1C). This reduction may be associated with the neuro‐pathological mechanisms of PTSD, such as abnormal neural circuits or impaired synaptic plasticity. Conversely, we observed an increasing trend in functional connectivity strength between some brain regions in the PTSD model group, particularly between the hippocampal CA1, CA2, and DG regions and the amygdala and cortical areas (olfactory, primary somatosensory) (p < 0.05, F (1, 88) = 12.69; Figure 1C). The average functional connectivity strength between brain regions of each group is shown in Figure 1D. The results showed that, between the two groups of mice, the hippocampus exhibited the strongest correlation with other brain regions (Figure 1D), highlighting its dominant role in these connectivity changes.

*The involvement of hippocampal ASTs in SPS‐induced PTSD‐like behaviors. (A) Analysis of the functional connectivity network characteristics of the hippocampus; (B) scanning each sampling volume (corresponding to four slices) in the height direction of 1.6 mm within 2.2 s; (C) brain regions with mean values from both groups and where there was a difference in functional connectivity (filtering condition: difference p < 0.05, correlation ecoefficiency > 0.15) (n = 3/group); (D) regional functional intensity (absolute difference between the average values of control group and model group mice); the involved brain regions include dentate gyrus (DG), thalamus (Tha), olfactory areas (Olf), perirhinal area (Per), striatum dorsal region (DS), visceral area (Vis), field CA1 (CA1), striatum‐like amygdalar nuclei (Str‐amy), hippocampal region (Hip), primary somatosensory area (SI), hypothalamic medial zone (Hyp), retrohippocampal region (Retro‐Hip), pallidum, dorsal region (Dor), primary somatosensory area (Pos), retrosplenial area (Rsp), perirhinal area (Per), hypothalamic lateral zone (Hyp), supplemental somatosensory area (SMA), agranular insular area (Agr), posterior parietal association area (Pos), temporal association areas (Tem), auditory areas (Aud); (E) diagram showing the number of astrocytes in local brain regions of mice; (F) statistical analysis of the differences in the number of astrocytes in various brain regions; (G) representative images of S100A10 immunofluorescence staining in the CA1 regions of the control group and SPS group (scale bar: 50 µm); (H) percentage of colabeling of S100A10 and GFAP in the control group and SPS group (n = 3/group); (I) statistical values of the average fluorescence intensity of S100A10 in the control group and SPS group. Data are presented as mean and SEM. p < 0.05 between groups (t‐test).

Subsequently, we performed whole‐brain GFAP immunofluorescence staining (Figure 1E). Compared with the control group, the number of GFAP‐positive ASTs in the hippocampus of the SPS group was significantly reduced (p < 0.05, t = 7.970, df = 104.0; Figure 1F). Additionally, in the hippocampal CA1 region of the SPS group, the colabeling rate of S100A10 and GFAP (p < 0.05, t = 17, df = 4; Figure 1G,H) and the mean fluorescence intensity of S100A10 (p < 0.05, t = 20.96, df = 4; Figure 1G,I) were significantly decreased compared with the control group. These results reveal that ASTs of the hippocampus might be the critical target cells in PTSD model mice.

Hippocampal AST Calcium Signals and ATP Release Were Reduced During Fear Memory Retrieval in PTSD Model Mice

2.2

Then, we used the gene‐encoded calcium indicator GCaMP6f in combination with an optical recording system to monitor calcium signals in hippocampal ASTs of freely moving mice in real‐time (Figure 2A). The specificity of GCaMP6f targeting AST was confirmed by immunofluorescence staining (Figure 2B). On the 14th day postmodeling, fiber optic recordings revealed that compared with control group mice, PTSD model mice exhibited a downward trend in the area under the curve (AUC) during the contextual fear memory test, though this difference was not statistically significant (p > 0.05, t = 1.096, df = 15; Figure 2C,D,E). However, the SPS group mice had exhibited significantly decreased peak number (p < 0.05, t = 3.068, df = 15; Figure 2F) and significantly increased freezing time during the first 5 min compared with the control group mice (p < 0.001, t = 5.838, df = 15; Figure 2G). Additionally, there was a significant negative correlation between the peak numbers of calcium signal and freezing time in the contextual fear test (p < 0.001, r = 0.6279; Figure 2H).

*Hippocampal AST calcium signals and ATP release were reduced during fear memory retrieval in PTSD model mice. (A) Flowchart of the experiment (n = 8/group); (B) representative graph of GcaMP6f‐labeled ASTs; (C) representative graph of the calcium signal changes in the hippocampus of mice during the contextual fear test; (D) difference graph of the calcium transient AUC of ASTs; (E) quantitative data of the calcium transient AUC of AST; (F) peak number of calcium signal releases in the contextual fear test of mice; (G) freezing time of mice in the contextual fear test; (H) correlation analysis between the peak number of calcium signal and the freezing time of mice; (I) flowchart of the experiment (n = 6/group) and representative graph of ATP fluorescent probe virus‐labeled AST; (J) representative graph of the ATP fluorescence signal changes in the hippocampus of mice during the contextual fear test; (K) difference graph of ATP fluorescence signal AUC; (L) quantitative data of ATP fluorescence signal AUC; (M) peak mean of ATP fluorescence signal release; (N) peak number of ATP fluorescence signal releases; (O) freezing time of mice in the contextual fear test. Data are presented as mean and SEM. *p < 0.05, **p < 0.01, **p < 0.001 between groups (t‐test).

We further injected an ATP fluorescent probe into the hippocampal region of mice and allowed 2 weeks for expression before inducing the PTSD model using the SPS method. One week later, optical fibers were implanted, and an optical fiber recording system was used to monitor extracellular ATP levels during the contextual fear memory test in real‐time (Figure 2I). Immunofluorescence staining confirmed high colocalization of the ATP probe virus with ASTs (Figure 2I). Fiber optic recordings revealed that during the contextual fear test, the SPS group mice exhibited significantly lower AUC (p < 0.05, t = 2.331, df = 10; Figure 2J,K,L), peak mean (p < 0.05, t = 2.663, df = 10; Figure 2M), and peak number (p < 0.05, t = 2.306, df = 10; Figure 2N) of ATP fluorescence signal compared with the control group. Additionally, the freezing time during the contextual fear test was significantly increased in SPS group mice (p < 0.001, t = 8.950, df = 15; Figure 2O). These results indicate that hippocampal AST calcium signals and ATP release of PTSD model mice were reduced during fear memory retrieval.

Reduction of AST PGC‐1α–CX43 Axis in SPS Mice Was Responsible for the Enhanced Fear Retrieval

2.3

Moreover, given that our previous research revealed a significant structural remodeling of ASTs in the hippocampus of SPS mice, characterized by a marked reduction in branch number and decreased PGC‐1α expression [14], we performed high‐purity sorting of ASTs using fluorescence‐activated cell sorting (FACS) to analyze the cellular source of PGC‐1α downregulation in the SPS model (Figure 3A,B). Western blot analysis of the sorted ASTs revealed that compared with the control group, the SPS group mice exhibited significantly decreased PGC‐1α expression in ASTs (p < 0.001, t = 9.972, df = 6; Figure 3C) and a significant reduction in the expression of the downstream signaling molecule nuclear respiratory factor 1 (NRF1) (p < 0.01, t = 9.056, df = 6; Figure 3C). Meanwhile, we investigated whether CX43 expression changed in the hippocampus of SPS mice and the results showed a significant downregulation of CX43 expression on the 14th day post‐SPS modeling (p < 0.05, F (3, 12) = 6.062; Figure 3D). Additionally, we conducted related studies in female animals. SPS was also able to induce PTSD‐like behaviors in female mice, and the alteration trends of the hippocampal PGC‐1α–CX43 axis remained consistent with those observed in male animals (Figure S1).

*Reduction of AST PGC‐1α–CX43 axis in SPS mice was responsible for the enhanced fear retrieval. (A) Experimental steps of FACS (n = 3/group); (B) gate strategy and sorting results; (C) representative bands and quantitative data of PGC‐1α and NRF1 proteins; (D) representative bands and quantitative data of hippocampal CX43 protein in the control group and on Days 1, 7, and 14 after SPS modeling; (E) flowchart of the experiment (n = 6/group); (F and G) freezing time of Con+CSF and Con+GAP27 group mice in the contextual fear test (F) and the cued fear test (G); (H) ELISA for detecting the ATP levels in the hippocampal tissues of the mice in the Con+CSF and Con+GAP27 groups; (I) ELISA for detecting the ATP levels in the hippocampal tissues of the mice in the control and SPS groups. Data are presented as mean and SEM. *p < 0.05, **p < 0.001 between groups (t‐test or Tukey test).

To this end, we further examined the effects of CX43 blockade on fear memory and extracellular ATP levels in the hippocampus. After implanting injection cannulas into the CA1 region of the mouse hippocampus and allowing 1 week for recovery, fear conditioning response (FCR) tests were conducted (Figure 3E). The results showed that intracerebral microinjection of the CX43 mimetic peptide Gap27 (100 µM) significantly increased the freezing time of mice in the contextual fear test compared with the Con+CSF group (p < 0.05, t = 6.836, df = 10; Figure 3F), but not cued fear test (Figure 3G). Enzyme‐linked immunosorbent assay (ELISA) results showed that the blockade of CX43 reduced ATP levels in the hippocampus (Figure 3H). Additionally, ELISA results showed that ATP levels of the hippocampus were significantly decreased in the SPS mice compared with the control group mice (p < 0.05, t = 5.800, df = 10; Figure 3I). These results indicate that the reduced extracellular ATP levels in ASTs during fear memory retrieval in SPS mice may be associated with the decreased CX43 expression in the hippocampus.

AST PGC‐1α KD Enhanced Fear Memory

2.4

To elucidate the role of PGC‐1α in ASTs in regulating fear memory and anxiety‐like behaviors, we knocked down AST PGC‐1α in the hippocampus using AAV‐mediated RNA interference (Figure 4A). Four weeks postviral infection, immunofluorescence staining confirmed specific expression of the PGC‐1α KD virus in ASTs (colocalization with GFAP^+^ > 90%) (Figure 4B). Western blot analysis showed that PGC‐1α expression in the hippocampus of PGC‐1α KD group was significantly reduced compared with the scramble group (p < 0.001, t = 7.020, df = 6; Figure 4C). Concurrently, CX43 expression was also downregulated after PGC‐1α KD (p < 0.001, t = 4.616, df = 6; Figure 4C). These findings suggest that PGC‐1α may regulate AST network communication via CX43.

*AST PGC‐1α knockdown enhanced fear memory and anxiety‐like behaviors and impaired AST structure. (A) Flowchart of the experiment (n = 9/group); (B) representative images of virus‐labeled astrocytes; (C) representative bands and quantitative data of PGC‐1α and CX43; (D) freezing of mice in the contextual fear test; (E) freezing of mice in the cued fear test; (F) the percentage of time that mice spent exploring the open arm in the elevated maze test out of the total time; (G and H) the total distance and representative traces of the open field test; (I and J) the results and representative images of the Sholl analysis of astrocytes. The data are presented as mean ± standard error (n = 3/group). (K) Representative image of S100A10 immunofluorescence staining (scale bar: 50 µm); (L) the percentage of colabeling of S100A10 and PGC‐1α (n = 3/group). Data are presented as mean and SEM. *p < 0.05, **p < 0.01, **p < 0.001 between groups (t‐test).

Behavioral results showed that mice in the PGC‐1α KD group exhibited significantly longer freezing times in the contextual fear test compared with the scramble group (p < 0.001, t = 9.267, df = 16; Figure 4D). Similarly, in the cued fear test, the freezing time of PGC‐1α KD mice was significantly increased compared with the scramble group (p < 0.01, t = 3.496, df = 16; Figure 4E). Elevated plus maze (EPM) test revealed that PGC‐1α KD mice spent less time in the open arm than the scramble group (p < 0.05, t = 2.135, df = 16; Figure 4F). Open field test (OFT) showed that there was no statistically significant difference in the total distance between the PGC‐1α KD mice and the scramble group (p > 0.05, t = 0.3452, df = 16; Figure 4G,H), which suggested that PGC‐1α KD did not affect the basic motor function of mice. Besides, Sholl analysis demonstrated a significant reduction in the number of ASTs in the CA1 region of the hippocampus in PGC‐1α KD mice, particularly at branches 30–60 µm from the cell body (p < 0.05, F (1, 137) = 29.45; Figure 4I,J), and those data were consistent with previous report of structural changes in ASTs due to SPS modeling [16]. In addition, reactive ASTs are categorized into A1 and A2 types, with S100A10 serving as a marker for neuroprotective AST A2 type [20]. We evaluated S100A10 expression in the hippocampal CA1 region of scramble and PGC‐1α KD mice. Results showed that S100A10 expression was downregulated in the PGC‐1α KD group compared with the scramble group (p < 0.05, t = 7, df = 4; Figure 4K,L). These findings indicate that PGC‐1α in ASTs plays a crucial role in regulating fear memory retrieval and maintaining AST morphological complexity.

AST PGC‐1α KD Impaired Hippocampal Neuronal Activity

2.5

To evaluate neuronal activity in the hippocampal CA1 region during the FCR test, we analyzed spike signals from the hippocampal CA1 region of awake mice using in vivo electrophysiological recording (Figure 5A). Immunofluorescence staining showed high colocalization between the PGC‐1α KD virus and the AST marker GFAP, indicating specific viral expression in ASTs (Figure 5B). Multichannel in vivo electrophysiological recordings revealed that on the first day of fear memory training (baseline), there were no significant differences in mean firing frequency (p > 0.05, F (1, 120) = 10.91; Figure 5C) or burst frequency (p > 0.05, F (1, 120) = 17.26; Figure 5D) between the PGC‐1α KD and scramble groups. However, the asymmetry index (AI) was significantly higher in the PGC‐1α KD group (p < 0.05, F (1, 120) = 10.99; Figure 5E). During the contextual and cued fear tests, the firing frequency of PGC‐1α KD mice was significantly lower than that of the scramble group (p < 0.05, F (1, 120) = 10.91; Figure 5C,F; p < 0.05, F (1, 120) = 17.26; Figure 5D,F), while the AI value in PGC‐1α KD group did not show significant difference compared with the scramble group (p > 0.05, F (1, 120) = 10.99; Figure 5E,F).

*AST PGC‐1α knockdown impaired hippocampal neuronal activity. (A) Flowchart of the experiment; (B) representative images showing the colocalization of the fluorescence of the PGC‐1α knockdown virus (green) with the immunofluorescence of the AST marker GFAP (red) and the electrode implantation location (scale bar: 50 µm); (C–E) mean frequency (C), bursts per minute (D), and AI value (E) of mouse hippocampal neurons recorded by multichannel electrodes during the conditioned fear test (n = 5/group); (F) grating diagram of mouse hippocampal neurons spike; (G) schematic of the patch‐clamp experiment (n = 5/group); (H) original current representation of action potential; (I) statistical data of action potential; (J) original current representation of spontaneous excitatory postsynaptic current; (K) frequency of spontaneous excitatory postsynaptic current; (L) amplitude of spontaneous excitatory postsynaptic current. Data are presented as mean and SEM. *p < 0.05, *p < 0.01 between groups (t‐test).

We further investigated the influence of AST PGC‐1α KD on neurons in the hippocampal CA1 region using whole‐cell patch‐clamp recordings (Figure 5G). Compared with the scramble group, the PGC‐1α KD group exhibited decreased neuronal firing rate of action potential (AP) (p < 0.05, F (1, 520) = 37.46; Figure 5H,I) and reduced frequency of spontaneous excitatory postsynaptic currents (sEPSC) (p < 0.05, t = 2.262, df = 42; Figure 5J,K), while sEPSC amplitude remained unchanged (p > 0.05, t = 1.571, df = 41; Figure 5J,L). These results illuminate that AST PGC‐1α KD may reduce the neuronal activity in the hippocampal CA1 region.

AST PGC‐1α KD Reduced the Release of Extracellular ATP

2.6

Next, we detected the ATP release around the PGC‐1α KD ASTs using ATP probe virus. Immunofluorescence staining revealed that the PGC‐1α KD virus was specifically expressed in ASTs, with high colocalization between PGC‐1α and the ATP probe virus within these cells (Figure 6A,B). During the initial training and adaptation phase of fear memory, there was no significant difference in ATP levels between the scramble and PGC‐1α KD groups (p > 0.05, t = 0.564, df = 6; Figure 6C). However, in the contextual fear test, ATP levels in the PGC‐1α KD group were significantly lower than those in the scramble group (p < 0.05, t = 2.523, df = 6; Figure 6D,F,G). After excluding events with ΔF/F values below 2%, the peak mean (p < 0.05, t = 2.641, df = 6; Figure 6H) and peak numbers of ATP release during the contextual fear test (p < 0.05, t = 2.528, df = 6; Figure 6I) were higher in the scramble group compared with the PGC‐1α KD group. In contrast, no significant difference in ATP levels was observed between the groups during the cued fear memory test (p > 0.05; Figure 6E,J–M). These results suggest that the reduced extracellular ATP levels may contribute to enhanced fear memory retrieval in the PGC‐1α KD group.

*AST PGC‐1α knockdown prohibited the release of extracellular ATP. (A) Flowchart of the experiment (n = 4/group); (B) representative images of virus‐labeled astrocytes (scale bar: 50 µm); (C) representative graphs and quantitative data of the area under curve (AUC) of ATP fluorescence signal during the first training period; (D and E) representative graphs showing the changes in ATP levels in the two groups of mice during the contextual fear memory (D) and cued fear memory (E) tests; (F) representative graphs and heat maps of the AUC of ATP levels during the contextual fear memory test; (G) quantitative data of the AUC of ATP levels during the contextual fear memory test; (H and I) peak values (H) and frequencies (I) of ATP release during the contextual fear memory test; (J) representative graph and heatmap of the AUC of ATP levels in the two groups of mice during the fear memory test; (K) quantitative data of the AUC of ATP levels during the fear memory test; (L and M) peak values (L) and frequency (M) of ATP release during the fear memory test. Data are presented as mean and SEM. p < 0.05 between groups (t‐test).

Activation of Hippocampal AST or PGC‐1α Reduced Fear Memory Retrieval in SPS Mice

2.7

We injected an adeno‐associated virus (AAV8–‐GFAP–hM3Dq–mCherry) encoding hM3Dq into the hippocampal CA1 region to observe whether activation of hippocampal AST could reduce the fear memory retrieval of SPS mice (Figure 7A). Immunofluorescence staining revealed high colocalization of mCherry with the AST marker GFAP, while the neuronal marker NeuN did not show overlap with hM3Dq expression, indicating that hM3Dq was almost exclusively expressed in ASTs (Figure 7B). For FCR test, we administered CNO immediately after the first day of training. Results showed that compared with the SPS group mice receiving normal saline, CNO injection to activate ASTs significantly improved contextual fear memory in SPS mice (p < 0.05, t = 7.337, df = 12; Figure 7C,D). Additionally, we investigated whether AST activation could improve anxiety‐like behaviors in SPS mice using the EPM test. The results showed an upward trend of the percentage of time SPS mice spent in the open arm without statistically significant (p > 0.05, t = 1.538, df = 12; Figure 7E).

*Activation of hippocampal AST or PGC‐1α reduced fear memory retrieval in SPS mice. (A) Flowchart of experiment 5 (n = 7/group); (B) representative image of immunofluorescence of hM3Dq‐labeled AST; (C) freezing time of mice in the contextual fear test; (D) freezing time of mice in the cued fear test; (E) percentage of time that mice spent exploring the open arm in the elevated maze test out of the total time; (F) flowchart of the experiment 5 (n = 6/group); (G) freezing time of mice in the contextual fear memory; (H) freezing time of mice in the cued fear memory; (I) percentage of time that mice spent exploring the open arm in the elevated maze test out of the total time; (J and K) representative bands of GFAP (J) and CX43 (K) along with quantitative data; (L) flowchart of the experiment 6 (n = 7/group); (M) schematic diagram of the hM3Dq–mCherry/AAV5–PGC‐1α viral injection site and colabeling; (N) freezing time of mice in the contextual fear memory; (O) freezing time of mice in the cued fear memory. Data are presented as mean and SEM. *p < 0.05, **p < 0.01, **p < 0.001 between groups (t‐test or Tukey test).

Given the downregulation of PGC‐1α, we also investigated whether activation of PGC‐1α signaling could improve fear memory retrieval and anxiety‐like behaviors in SPS model mice using the agonists of PGC‐1α ZLN005 and resveratrol (Figure 7F). Behavioral tests showed that both ZLN005 and resveratrol significantly reduced freezing time in SPS mice during contextual (p < 0.01, F (3, 20) = 23.13; Figure 7G) and cued (p < 0.001, F (3, 20) = 11.11; Figure 7H) fear memory tests. In the EPM test, both agonists increased the percentage of time mice spent in the open arm (p < 0.01, F (3, 20) = 10.79; Figure 7I). We further examined the effects of PGC‐1α activation on the gap junction protein CX43 in ASTs. Western blot analysis revealed decreased expression of GFAP and CX43 in the hippocampal tissues of SPS mice, which was restored by ZLN005 and resveratrol treatment (GFAP: p < 0.05, F (3, 12) = 1.849; CX43: p < 0.05, F (3, 12) = 3.060; Figure 7J,K), suggesting that PGC‐1α pathway agonists may inhibit fear retrieval in SPS mice by restoring AST function.

To further verify the role of PGC‐1α in ASTs in fear memory, we simultaneously injected PGC‐1α KD and AAV8–GFAP–hM3D(Gq)–mCherry viruses into the hippocampal CA1 region of mice, followed by contextual and cued behavioral tests (Figure 7L). Colabeling of AAV5–PGC‐1α with hM3Dq–mCherry was shown in Figure 7M. The results showed that when PGC‐1α was knocked down, rescue effect of chemogenetic AST activation on fear retrieval in either the contextual (p < 0.001, t = 6.848, df = 12; Figure 7N) or cued (p < 0.01, t = 3.850, df = 12; Figure 7O) tests were eliminated.

Discussion

3

In this study, we found that PGC‐1α in hippocampal ASTs was downregulated in PTSD model mice. Moreover, interfering with PGC‐1α expression in hippocampal ASTs inhibited CX43, reduced ATP release, and downregulated the excitability of hippocampal neurons, thereby promoting fear memory retrieval. Additionally, activation of hippocampal AST or PGC‐1α reduced fear memory retrieval in SPS mice. Especially, PGC‐1α KD attenuated the effects of chemogenetic AST activation on fear retrieval in PTSD mice. These results identify astrocytic PGC‐1α a critical mediator of aberrant fear memory in PTSD, thus implicating it as a potential therapeutic target.

PGC‐1α‐Dependent Hippocampal AST–Neuron Coupling is Required for Fear Memory Retrieval

3.1

ASTs are among the most abundant glial cells in the brain and play a crucial role in neural function, including the regulation of neuronal development and synaptic plasticity [8, 21]. They are also closely associated with neuronal repair and the pathogenesis of various nervous system diseases. Notably, the role of ASTs in PTSD has garnered significant attention. First, ASTs and the loss of their function are considered key mechanisms in PTSD development [7]. Second, pharmacological study has shown that the recovery of AST function is involved in the protection of PTSD symptoms [22]. Our previous research found that SPS‐induced PTSD model mice exhibited enhanced fear memory retrieval and significant structural changes in ASTs, including reduced protrusion length and number [10]. Other study has demonstrated that AST activation can downregulate fear memory retrieval in fear conditioning models [9]. Additionally, ASTs in the hippocampal CA1 region receive cholinergic input from the posterior basal forebrain to regulate fear memory extinction, further indicating their potential involvement in PTSD [23]. Based on these findings, we investigated the effects of PGC‐1α KD in hippocampal ASTs on neuronal and behavioral functions. Intriguingly, PGC‐1α KD in hippocampal ASTs promoted fear memory retrieval and induced anxiety‐like behaviors in mice. PGC‐1α is a key transcriptional coactivator regulating mitochondrial biogenesis, energy metabolism, and antioxidant defense mechanisms. In our study, PGC‐1α KD in ASTs affected CX43 expression, decreased AST protrusion number and length, and reduced the number of A2‐type ASTs. These results suggest that PGC‐1α is critical in determining AST function.

ASTs account for 20% of total brain oxygen consumption to generates ATP [24], primarily through oxidative phosphorylation in AST mitochondria [25]. Recent studies have shown that dysregulation of extracellular ATP is linked to depression, and AST‐derived ATP can improve depression‐like behaviors [26, 27]. However, the molecular mechanisms by which AST‐released ATP regulates PTSD‐like behavior remain largely unknown. In this study, we found that PGC‐1α KD in ASTs reduced ATP release during fear memory retrieval. CX43, the major connexin protein in ASTs, is a key Ca^2^ ^+^‐dependent mechanism for ATP release [28, 29]. We observed a significant decrease in CX43 level in the hippocampus of PGC‐1α KD mice, indicating that PGC‐1α can modulate CX43 levels. Furthermore, blocking CX43 inhibited hippocampal ATP levels and enhanced fear memory in control mice. Based on these findings, we propose that PGC‐1α is crucial for maintaining AST activity, potentially through regulating CX43 levels to modulate ATP release. Given the pivotal role of PGC‐1α in regulating mitochondrial function, we also examined the expression of NRF1, a key transcription factor that regulates mitochondrial function [12]. Our results showed that the expression patterns of NRF1 and PGC‐1α were consistent in the hippocampus of PTSD model mice. This finding also suggests that PGC‐1α likely influences ATP release by modulating mitochondrial function. Interestingly, ATP release is also regulated by CX43. On one hand, CX43 can function as a hemichannel to mediate the release of ATP from the intracellular space to the extracellular environment. On the other hand, it facilitates the direct intercellular transfer of ATP via gap junctions. In future, directly monitoring mitochondrial dynamics or elucidating the specific transport pathways of ATP would contribute to a deeper understanding of the pathological mechanisms underlying PTSD.

Increased fear memory in PTSD is associated with heightened amygdala activity, which is negatively regulated by the hippocampus [30]. Hippocampal dysfunction in PTSD patients may lead to a loss of inhibitory control over the amygdala, resulting in persistent fear responses. In our model, PGC‐1α KD in ASTs may impair hippocampal function, thereby disrupting its inhibitory effect on the amygdala. Direct synaptic connections between ASTs are crucial for regulating neuronal function. While we cannot rule out the possibility of direct AST regulation of the amygdala, our in vivo electrophysiological and in vitro patch‐clamp recordings demonstrated that PGC‐1α KD in the CA1 region reduced hippocampal neuronal firing rates and burst numbers, indicating an increased AP threshold in hippocampal neurons. Furthermore, patch‐clamp analysis revealed that AST‐specific PGC‐1α KD selectively reduced the frequency of CA1 neuronal sEPSCs without affecting their amplitude, suggesting a predominant impairment of presynaptic function in hippocampal neurons. Extensive reports have highlighted the regulatory effects of AST on neuronal and synaptic connections, influencing animal behavioral functions [31, 32]. Abnormal AST–neuron coupling is also recognized as a crucial mechanism underlying diseases such as depression [33]. Our current research further suggests that impaired AST–neuron coupling may be a key factor in aberrant fear memory retrieval, contributing to the development of PTSD. However, how the SPS model transmits signals to ASTs and subsequently disrupts presynaptic function remains to be elucidated and warrants further investigation. In future studies, we will employ conditional knockout of PGC‐1α, combined with spatial transcriptomics and neurobiological techniques, to elucidate the role of AST–neuron coupling in fear memory.

Hippocampal AST PGC‐1α is an Important Target for PTSD Intervention

3.2

PTSD is a complex disease, involving multiple brain regions and multiple mechanisms in the occurrence of its symptoms [30]. To a certain extent, PTSD is regarded as a disorder involving dysregulation of fear memory [30], which is generally believed to be caused by functional disorders of brain neurons or neural circuits, and thus research on PTSD is more focused on neurons or neural circuits [34, 35]. However, an increasing number of experimental data suggest that AST can also actively regulate synaptic transmission and plasticity, and eventually participate in the processing of fear memory [36, 37]. In this study, we analyzed the changes in cerebral blood flow in various brain regions of SPS mice and established the corresponding connections [38]. Our results showed that the connections between multiple brain regions centered on the hippocampus were altered in SPS mice. This finding further indicates the crucial role of the hippocampus in the occurrence and treatment of PTSD [39, 40, 41]. Additionally, we used the PanoBrain scanning microscope to obtain GFAP staining images with a full field of view. The results showed that the number of GFAP‐positive cells was most prominent in the hippocampal region. Interestingly, in the SPS model mice, the number of GFAP‐positive cells in the hippocampus region decreased, while the number of GFAP‐positive cells in the amygdala region increased. These results indicate that the decrease in AST in the hippocampal region may be an important cause for the increased fear memory in PTSD. Consistent with our findings, Gill et al. also pointed out through positron emission tomography detection methods that AST loss rather than AST proliferation occurred in the brains of PTSD patients [42].

Unlike neurons, in the absence of obvious electrical signals, the excitability of AST mainly relies on highly spatio‐temporally coordinated fluctuations of intracellular Ca^2+^ concentrations [43]. In this study, we performed a real‐time monitor on the Ca^2+^ dynamics of AST in PTSD model mice during the process of fear memory retrieval and found that the AST in the hippocampal region of SPS mice exhibited abnormal Ca^2+^ signal transduction during the retrieval of fear memory which was consistent with the behavioral changes of the mice, that is, the peak number of calcium signal release in the mice was correlated with the freezing time. It is noteworthy that by specifically manipulating the Ca^2+^ signaling in AST through chemical genetics (Gq‐DREADDS) in vivo, the fear memory retrieval in SPS mice can be improved. Interestingly, during memory retrieval, the AUC of calcium signals in PTSD model mice did not show a significant decrease compared with the control group. This observation may be related to the time window selected for signal analysis. Additionally, we noted potential differences in the kinetics of calcium signals between the model and control groups, which warrants further investigation.

PGC‐1α is a well‐known coactivator involved in mitochondrial function [44] and is also associated with the pathophysiology of multiple neurological disorders. Furthermore, many of its functions are attributed to specific cell types within the central nervous system [45, 46]. In the central nervous system of mice, PGC‐1α is involved in maintaining the gene expression programs for synchronous neurotransmitter release, structure, and metabolism in excitatory neurons of the cortex and hippocampus [47]. Additionally, the deficiency of PGC‐1α disrupts the vitality and function of interneurons and dopaminergic neurons, subsequently leading to changes in inhibitory signal transduction and behavioral dysfunction [47, 48]. During the development of AST, PGC‐1α exerts a great influence on regulating the generation of mitochondria and inhibiting the proliferation and maturation of AST after birth. Abnormal maturation of AST can also reduce the formation of excitatory synapses [16]. In this study, we found that interfering with the expression of PGC‐1a in mature ASTs would affect the structure of ASTs. We also discovered that SPS modeling would lead to a significant downregulation of PGC‐1α expression in hippocampal ASTs of mice, and specifically interfering with the expression of PGC‐1α in hippocampal ASTs could increase the fear memory of mice and inhibit the expression of gap junction protein CX43. Furthermore, the low expression of PGC‐1α in AST will affect the extracellular ATP level. Therefore, we hypothesize that the low expression of PGC‐1α in AST may influence ATP release by interfering with the coupling between AST, thereby affecting the behavioral functions of mice.

In order to further explore whether PGC‐1α can serve as a therapeutic target for PTSD, we continuously administered the PGC‐1α‐specific activator ZLN005 and the SIRT1 agonist resveratrol in SPS mice [49]. We found that the abnormal fear memory and anxiety behaviors of SPS mice were improved after PGC‐1α activation. Further experiments demonstrated that continuous administration of ZLN005 and resveratrol could improve the decreased expression of GFAP and CX43 in the hippocampus of SPS mice. In this study, we believed that the long‐term activation of PGC‐1α might induce the expression of upstream molecules to compensate for the damage to this signaling pathway. As it was, we chose to administer ZLN005 multiple times rather than in a single dose. Although this study has clarified the therapeutic potential of PGC‐1α agonists, and existing research in the field has confirmed their protective effect on neurological diseases [50]. However, the penetration of PGC‐1α agonists in the blood–brain barrier still needs to be further improved. In the future, delivery strategies such as nanocarriers and focused ultrasound technology can be considered to enhance delivery. In addition, since PGC‐1α is also widely expressed in peripheral tissues (liver, muscle), nonspecific activation may lead to side effects such as metabolic disorders. Subsequent studies will focus on the optimization of brain‐targeted delivery systems and tissue‐specific gene editing strategies, laying the foundation for the clinical transformation of PGC‐1α agonists.

Limitations of the Study

3.3

Through this study, we demonstrate that PGC‐1α deficiency in hippocampal ASTs underlies enhanced fear memory retrieval in PTSD model mice. However, the mechanism leading to the decrease of PGC‐1α in ASTs during PTSD progression remains unclear. In subsequent research, we will further investigate the regulatory mechanisms of PGC‐1α downregulation from an epigenetic perspective. Additionally, while the primary function of PGC‐1α is to regulate mitochondrial function, its specific modulatory effect on the gap junction protein CX43 requires further validation. Moreover, although our study indicates that SPS can induce PTSD‐like behaviors in female mice, with trends in PGC‐1α and CX43 changes consistent with those in males, whether the underlying mechanisms are identical remains to be elucidated.

Conclusion

4

This study demonstrates that the downregulation of PGC‐1α in ASTs is a key factor contributing to the enhanced fear memory retrieval in PTSD by weakening AST–neuron coupling. These findings suggest that hippocampal PGC‐1α is a critical target for PTSD intervention.

Materials and Methods

5

Animals and Ethics Statement

5.1

Male and female C57BL/6J mice (2 months old; 20–25 g) were purchased from Hangzhou Ziyuan Laboratory Animal Technology Co. (Production License No.: SCXK (Zhe) 2019‐0004). The mice were housed in animal room with a temperature of 22 ± 2°C and a relative humidity of 45–65%, under a 12‐h light/dark cycle, with ad libitum access to drinking water and food. All experimental procedures were approved by the Ethics Committee of Anhui University of Chinese Medicine (AHUCM‐mouse‐2022014).

Model Preparation and Experimental Designs

5.2

The SPS regime was selected to induce mouse behaviors analogous to PTSD [51, 52, 53]. First, mice were confined in a restraining tube for 2 h. After allowing the mice to acclimatize, they were immediately placed in a transparent acrylic cylindrical container, where they underwent forced swimming for 20 min. The mice were returned to their cages for rest for 15 min. Subsequently, the mice were exposed to an isoflurane environment to induce anesthesia until they lost consciousness (within <1 min), a procedure that was repeated twice. After a 15‐min rest, the mice received a single electrical shock (2 mA, 2 s). After the modeling was completed, the mice were returned to their cages for quiet housing. To reduce nonspecific interference, the control group mice received similar handling, but did not receive SPS regime.

To achieve the research objectives, this study designed five experiments. In Experiment 1, mice were randomly divided into a control group and a model group (n = 3/group). fUS imaging and light‐sheet microscopy were used to assess the strength of functional connectivity between brain regions and the pan‐expression of GFAP in the brain. In Experiment 2, AST calcium (n = 8/group) and ATP release (n = 6/group) dynamics of PTSD model mice during fear memory retrieval were recorded using fiber photometry. In Experiment 3 (n = 3/group), ASTs were sorted using FACS, and Western blot was employed to detect the expression of PGC‐1α in PTSD model mice. In addition, the effects of CX43 gap junction blocker on fear memory retrieval and ATP release were observed (n = 6/group). In Experiment 4, an AAV5–GfaABC1D–EGFP–shPGC‐1α virus was designed. Behavioral tests, in vivo electrophysiology, whole‐cell patch‐clamp, and extracellular ATP level measurements were conducted to evaluate the effects of PGC‐1α KD. In Experiment 5, the effects of chemogenetics activation of ASTs or pharmacological activation of PGC‐1α on fear memory in PTSD model mice were assessed. In Experiment 6, the effects on fear memory in a PTSD mouse model were assessed after hippocampal PGC‐1α KD via viral delivery, followed by chemogenetic activation of ASTs or pharmacological activation of PGC‐1α.

Drug Administrations

5.3

Clozapine N‐oxide (CNO; BrainVTA) was dissolved in saline, and the control group was injected with an equivalent volume of saline. CNO (3 mg/kg) was administered intraperitoneally 30 min before the contextual fear test [54]. The PGC‐1α activator ZLN005 (HY‐17538; MCE) was administered at a dose of 15 mg/kg/day [55, 56], while the SIRT1 activator resveratrol (HY‐16561; MCE) was administered at a dose of 40 mg/kg/day once a day for 14 days [57, 58]. ZLN005 and resveratrol were dissolved in a vehicle solution containing 5% DMSO, 40% PEG300, 5% Tween‐80, and 50% saline. Gap 27, a CX43 gap junction blocker (HY‐P0139; MCE), was intracranially administered at 0.5 µL per side with a concentration of 100 µM [59].

Following anesthesia in an induction chamber, mice were positioned in a stereotaxic frame. A stainless‐steel double guide cannula (62060; RWD Life Science, Shenzhen) with an inner stylet (62231; RWD Life Science) was implanted above the hippocampal CA1 injection site (coordinates: bregma: −1.82 mm; L: ±1 mm; H: 1.50 mm). The cannula was secured with dental cement, and animals were returned to their home cages for a 7‐day recovery period. Microinjections were performed using a dummy cannula (62131; RWD Life Science), with either Gap27 dissolved in artificial cerebrospinal fluid (ACSF) or ACSF alone (0.5 µL) infused at 0.1 µL/min. The injection cannula was maintained in situ for 5 min postinfusion to minimize drug diffusion along the injection tract, followed by behavioral test 30 min later.

Viral and Stereotaxic Injection Protocol

5.4

Mice were anesthetized in an isoflurane induction chamber (4% isoflurane) and then secured in a stereotaxic frame (RWD Life Science), with isoflurane adjusted to 0.5–1%. After routine disinfection, the skull was exposed, and erythromycin ointment was applied to protect the eyes. GCaMP calcium indicator virus and ATP fluorescence probe virus were unilaterally injected into the hippocampus (bregma: −2.3 mm; L: 1.75 mm; H: 1.55 mm), with a volume of 500 nL. Chemogenetic viruses were bilaterally injected into the hippocampus using the same coordinates, with a volume of 500 nL. shPGC‐1α virus was bilaterally injected at two hippocampal sites (coordinate 1: bregma: −2.3 mm; L: 1.75 mm; H: 1.55 mm; coordinate 2: bregma: −3.08 mm; L: 3.10 mm; H: 3.50 mm), with 400 nL injected at coordinate 1 and 300 nL at coordinate 2.

The following viral vectors were used: AAV5–GfaABC1D–GcaMP6f (viral titer 5.85 × 1012 µg/mL; BrainVTA), AAV8–GFAP–hM3Dq–mCherry (viral titer 8.25 × 1013 µg/mL; WZ Biosciences Inc), rAAV–GfaABC1D–rATP1.0 (viral titer 6.50 × 1012 µg/mL; BrainVTA), AAV2/5–GFaABC1D–EGFP–5′‐miR‐30a–shRNA(PGC‐1a)–3′‐miR30a–WPREs (viral titer 5.88 × 1012 µg/mL; BrainVTA), AAV–GFaABC1D–EGFP–5′‐miR‐30a–shRNA(scramble)–3′‐miR30a–WPREs (viral titer 5.35 × 1012 µg/mL; BrainVTA). The interference sequence was 5′‐CCCATTTGAGAACAAGACTAT‐3′ (targeting mouse PGC‐1α; NM_001402987.1) [60, 61]. Intracranial injections were performed using a microinjection pump controller and borosilicate glass micropipettes at 50 nL/min. The needle was left in place for 10 min postinjection to allow viral diffusion. The skin was sutured, and mice were kept on a heating pad during recovery before being returned to their home cages. To ensure signal stability, all calcium signal recordings were initiated 4 weeks after GCaMP6f expression. For shRNA virus experiments, functional assays were conducted 4 weeks postinjection to ensure maximal KD efficiency. Notably, in coinjection experiments of ATP fluorescent probe virus and PGC‐1α virus, a sequential injection strategy was adopted: PGC‐1α virus was injected first, followed by the ATP fluorescent probe virus 2 weeks later (after PGC‐1α expression was established).

fUS Imaging

5.5

A head plate was surgically implanted in the mouse skull using precise techniques, beginning with isoflurane‐induced anesthesia and subcutaneous medetomidine injection, followed by fixation in a stereotaxic apparatus. After skull exposure, cleaning, and drilling for anchor screws, the head plate was secured with dental cement. Postsurgery, mice recovered on a heating pad while wearing a 3D‐printed protective cap. For ultrasound probe positioning, mice were placed in an imaging setup, and real‐time cerebral blood volume imaging was used to calibrate the probe. Following angiography, IcoStudio software registered the data to the Allen Mouse Brain Atlas. The Brain Positioning System guided probe alignment to target brain regions based on vascular landmarks.

Behavioral Assessment

5.6

FCR Test

5.6.1

The FCR test consisted of one training session and two test sessions. On Day 14 post‐SPS modeling, mice were placed in a fear conditioning chamber (242 mm (W) × 242 mm (D) × 300 mm (H)) for a 3‐min habituation period, followed by three cycles of tone (28 s, 1 kHz, 90 dB) paired with foot shock (2 s, 0.8 mA), with a 2‐min rest period afterward. Twenty‐four hours later, contextual fear memory was assessed by re‐exposing mice to the same chamber for 5 min without tone or shock, and freezing behavior was recorded. Two hours later, cued fear memory was tested in a novel context with a neutral tone cued auditory stimulus (80 dB, 4 kHz, sine wave) for 3 min, followed by a 60‐s chamber exposure. Freezing duration was analyzed using ANY‐Maze software.

Elevated Plus Maze

5.6.2

The EPM test was used to evaluate anxiety‐like behavior. The maze consisted of two open arms (50 mm (D) × 250 mm (W)) and two closed arms (50 mm (W) × 250 mm (D) × 50 mm (H)) elevated 50 cm above the ground. Mice were placed in the central area facing a closed arm and allowed to freely explore for 5 min. The time spent in the open arms relative to total exploration time was calculated as an index of anxiety, with longer open‐arm time indicating lower anxiety.

Open Field Test

5.6.3

The OFT is employed to assess anxiety‐like behavior in rodents. At the initiation of the test, each mouse is placed in the center of a cubic arena (400 mm (W) × 400 mm (H) × 400 mm (D)) and allowed to move freely for 5 min. The Super‐Maze software is utilized to record the duration of time spent in the central zone, the distance traveled within the central zone, and the total distance traversed by the mice. A reduction in the time and distance spent in the central zone is indicative of more severe anxiety‐like behavior.

Electrophysiology

5.7

In Vivo Multichannel Electrophysiology

5.7.1

Mice were anesthetized in an isoflurane induction chamber (4%), and their heads were fixed on a stereotaxic apparatus (RWD Life Science). The isoflurane concentration was adjusted to 0.5–1%. After routine disinfection, the skull was exposed and leveled. Using a handheld drill (78001; RWD Life Science) with a 0.5 mm round burr, three holes were drilled above the hippocampal CA1 region. Screws (0.1 mm diameter) were inserted into these holes for dental cement fixation, with the ground and reference wires of the multichannel electrode wrapped around the screw posterior to the lambda. The electrode implantation coordinates were: bregma: −2.3 mm; L: 1.75 mm; H: 1.55 mm. The electrode was secured to an electric microdrive using an electrode holder and slowly inserted at 0.5 mm/s. Upon nearing the target location, the electrode remained in place for 30 min. Medical adhesive was applied to the electrode insertion site, and dental cement was used to fix the electrode to the skull surface. After 1 week of recovery, electrophysiological signals during FCR test were recorded using Cheetah 6.4.2 (Neuralynx, USA), with spike signal frequency and sampling rates set at 300–5000 Hz and 30 kHz, respectively. Spike sorting was performed using SpikeSort 3D 2.5.4 (Neuralynx), employing the KlustaKwik clustering method to group waveforms with similar principal components. Neuronal activity in the hippocampal CA1 region was characterized using NeuroExplorer 5 (Nex Technologies, USA), calculating mean firing rate (mean frequency), burst frequency (bursts per min), and AI. Mean frequency evaluates neuronal excitability, burst frequency reflects clustered firing patterns, and AI—calculated as the mode interspike interval (mode ISI) divided by the mean interspike interval (mean ISI)—indicates the likelihood of burst firing (lower AI values suggest higher burst propensity) [62, 63, 64].

Whole‐Cell Patch‐Clamp

5.7.2

Following deep anesthesia, mice were transcardially perfused with 15 mL of ice‐cold cutting solution containing (in mM): 93 NMDG, 2.5 KCl, 93 HCl, 1.2 NaH_2_PO_4_, 10 HEPES, 30 NaHCO_3_, 25 glucose, 5 Na‐ascorbate, 3 Na pyruvate, 2 thiourea, 10 MgSO_4_·7H_2_O, and 0.5 CaCl_2_ (pH 7.3, 300–310 mOsm/L). The brain was rapidly extracted and immersed in preoxygenated (95% O_2_, 5% CO_2_, v/v) cold slicing solution. Coronal slices (300 µm) containing the dorsal hippocampal CA1 region were prepared using a vibratome (VT‐1200; Leica, Wetzlar, Germany) and transferred to a recovery chamber containing (in mM): 92 NaCl, 10 HEPES, 2.5 KCl, 1.2 NaH_2_PO_4_, 2 MgSO_4_·7H_2_O, 2 CaCl_2_, 30 NaHCO_3_, 3 Na pyruvate, 2 thiourea, and 25 glucose (pH 7.3, 300–310 mOsm/L when carbogenated with 95% O_2_ and 5% CO_2_) at 32°C. After incubation for at least 60 min, whole‐cell patch‐clamp recordings were performed. The dorsal CA1 region was first identified under low magnification (×5), and pyramidal neurons near fluorescently labeled ASTs were selected under high magnification (×60) using an upright microscope (BX50WI; Olympus, Japan). Recordings were conducted using an Axon 700B amplifier and Digidata 1550B digitizer (Axon Instruments). Borosilicate glass micropipettes (5–10 MΩ) were pulled on a horizontal puller (P‐97; Sutter Instruments, USA) and filled with internal solution as shown in a previous literature [65]. sEPSCs were recorded in voltage‐clamp mode at −70 mV, while current‐induced AP was elicited in current‐clamp mode by injecting depolarizing current steps (−20 to 200 pA). Data were analyzed using Clampfit 10.7 (Molecular Devices, USA) and Mini Analysis Program (Synaptosoft, USA).

In Vivo Calcium and ATP Recordings and Analysis

5.8

AAV5–GFAP–cyto‐GCaMP6f and ATP probe viruses were injected into one side of the hippocampus. After 2–3 weeks of viral expression, the SPS model was established. At 7–8 days postmodeling, a black ceramic ferrule (Ф1.25 mm/200 µm, 0.39NA; RWD Life Science) was implanted, with the ferrule positioned 50–100 µm above the viral injection site, and fixed using dental cement. After the cement solidified, the mice were transferred to cages for health monitoring. A tri‐color multichannel fiber photometry system (R820; RWD Life Science) was used to record fluorescence signals, with a 410 nm LED as the control, a 470 nm LED for excitation‐dependent green GCaMP fluorescence and ATP 1.0 probe, and a 560 nm LED for the red ATP 0.5 probe. Postdata acquisition, the analysis module of the system was used to process raw signals, collecting data by analyzing fluorescence changes (ΔF) relative to the initial baseline fluorescence (F) and observing signal changes (ΔF/F) corresponding to calcium transients and ATP probes, where each peak on the ΔF/F plot represents AST responses to external stimuli. Raw data were preprocessed, including smoothing (to remove excessive noise and highlight target peaks), baseline correction (to address fluorescence decline due to prolonged recording and LED thermal effects), and motion correction (using 410 nm as the control channel to fit 410 nm data with 470 nm data and eliminate background noise). Fluorescence data were then synchronized with behavioral data, and event‐related signal analysis was performed for plotting. Finally, intergroup comparisons of the data were conducted.

ELISA Assay

5.9

The ATP levels in the hippocampus were measured using an ELISA assay. Briefly, 30 mg of hippocampal tissue from each group of mice was placed in a 1.5 mL centrifuge the ATP assay kit (JL‐T0633‐96; Jianglai Biotechnology, Shanghai, China). The results were calculated to obtain the ATP levels in the hippocampus.

FACS of Mouse Hippocampal ASTs

5.10

To dissociate the hippocampal tissues from each group of mice into single cells, tissue dissociation solution (abs9482; absin) was added into the tissues in a certain proportion (1:100). The cells were resuspended in 300 µL HBSS containing 5% serum to prepare a single‐cell suspension. Following the manufacturer's instructions, 14 µL FcR Blocking Reagent (mouse, 130‐092‐575; Miltenyi) was added, and ASTs were labeled with the surface marker antibody anti‐ACSA‐2‐PE (mouse, 130‐123‐284; Miltenyi). The ACSA‐2^+^ cell population was sorted using a Cytoflex SRT flow cytometer and collected into a receiving tube. The sorted ASTs were lysed with cell lysis buffer, and total protein was extracted for subsequent Western blot experiments.

Western Blot

5.11

Hippocampal tissues were homogenized and lysed, and protein concentrations were measured using a BCA protein assay kit. Protein samples were separated by SDS‐PAGE (12% gel, 55 V/15 min for stacking gel and 110 V/75 min for separating gel) and transferred to a nitrocellulose membrane at 400 mA for 30–40 min. The membrane was blocked with 5% skim milk (in PBS containing 0.1% Tween‐20) for 2 h at room temperature and then incubated overnight at 4°C with primary antibodies: PGC‐1α (1:1000; ZEN BIO), NRF1 (1:1000; ZEN BIO), GFAP (1:1000; bioss), CX43 (1:1000; ZEN BIO), β‐actin (1:1000; CST), and GAPDH (1:1000; ZEN BIO). After washing three times with PBST (PBS containing 0.1% Tween‐20, 3 × 10 min), the membrane was incubated with horseradish peroxidase‐conjugated goat anti‐rabbit IgG or goat anti‐mouse IgG (1:10,000) for 2 h at room temperature. Protein bands were visualized using enhanced chemiluminescence reagent (Tanon), and quantitative analysis of the protein bands was performed using ImageJ software.

Immunofluorescence Staining and Sholl Analysis

5.12

The mouse brains were fixed in 4% paraformaldehyde for 24 h, and then dehydrated in 30% sucrose for 48 h at 4°C and sectioned into 20 µm slices using a cryostat. Sections were blocked for 1 h in 0.1 M PBS containing 10% goat serum and 0.3% Triton X‐100, followed by overnight incubation at 4°C with primary antibodies: GFAP (1:100; bioss), NeuN (1:100; CST), and S100A10 (1:200; Biosciences). After three washes in PBS (3 × 15 min), sections were incubated with FITC‐conjugated anti‐rabbit IgG (1:100) for 2 h at room temperature, counterstained with DAPI, and mounted. Images were acquired using an FV3000 Olympus confocal laser scanning microscope and analyzed with ImageJ software.

Whole‐field GFAP immunofluorescence images were obtained using a PanoBrain section scanner. Subsequent analyses were performed in Panolyzer software: initial registration of the brain section images with the Allen Mouse Brain Atlas was achieved using the DeepSlice method, followed by fine‐tuning with nonlinear registration to ensure accurate anatomical alignment; threshold segmentation algorithms were applied to preliminarily quantify ASTs, and manual corrections were made to obtain final cell statistics.

Sholl analysis was performed directly on bitmap images using the Fiji toolkit integrated into ImageJ. The plugin generated concentric circles spaced 10 µm apart, starting from the center of the DAPI signal to the end of the farthest process, and counted the number of intersections with each circle to quantify AST morphology and arborization. For each group, three mice were analyzed, and Sholl analysis was conducted on ASTs in the hippocampal CA1 region. Three sections per animal (15 cells per section) from the apical dendritic layer of CA1 were analyzed by different researchers, and the average values were calculated.

Statistical Analysis

5.13

Data processing was performed using the statistical software GraphPad Prism 9.0, with results presented as mean ± SEM, and statistical methods including independent samples t‐test, one‐way ANOVA, or two‐way ANOVA, followed by Tukey's post hoc test to establish statistical significance at *p < 0.05, **p < 0.01, and ***p < 0.001.

Author Contributions

GQZ, JJW, and MC designed the study and wrote the manuscript. JW, XYC, DKC, and SJY conducted the research and analyzed the data. All authors reviewed the results and approved the final version of the manuscript.

Funding

This study was supported by the Scientific Research Program of Anhui Provincial Department of Education (2024AH040137, 2024AH051044), Research Funds of Center for Xin'an Medicine and Modernization of Traditional Chinese Medicine of IHM (2023CXMMTCM013, 2023CXMMTCM021), National Natural Science Foundation of China (82404890), Excellent Funding for Academic and Scientific Research Activities for Academic and Technological Leaders in Anhui Province (2022D317), Chinese Medicine Prevention and Treatment of Mental Illness Research Team (2024zyky02), Open Fund of High‐level Key Discipline of Basic Theory of TCM of the State Administration of Traditional Chinese Medicine (ZYJCLLZD‐04), Key Laboratory of Xin'an Medicine (2024xayx11), and Basic‐Clinical Integration Program of Anhui University of Chinese Medicine (JCLCA2025003).

Ethics Statement

All animal experiments in this study were complied with National Institutes of Health (NIH) guidelines for the Care and Use of Laboratory Animals and approved by the Animal Ethics Committee of Anhui University of Chinese Medicine (AHUCM‐mouse‐2022014).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: mco270671‐sup‐0001‐SuppMat.docx.