Introduction

Post-traumatic stress disorder (PTSD) is a mental disorder triggered by traumatic and threatening experience such as natural disasters, sexual assault, or automobile accidents [1]. In general, most individuals who have experienced traumatic events exhibit mild and transient adjustment difficulties, and typically recover within days or weeks. However, some individuals experience long-term adaptation difficulties and eventually develop PTSD, exhibiting symptoms such as avoidance, emotional numbness, intrusive memory, long-term anxiety, and even suicidal behavior [2]. The lifetime prevalence of PTSD varies by social background and country of residence, ranging from 1.3 to 12.2% [3]. The pathogenic mechanisms underlying PTSD, however, has not yet been fully elucidated, resulting in the use of medicines that primarily target the symptoms rather than the underlying causes.

Over the past two decades, distinct brain regions including medial prefrontal cortex (mPFC), dorsal anterior cingulate cortex (dACC), hippocampus and amygdala have been identified as constituting the neural circuits that mediate the core clinical symptoms of PTSD [4]. For instance, compared with the control group, individuals with PTSD often exhibit heightened amygdala activity and diminished mPFC activity during symptom provocation studies. This suggests a failure of the cortex to inhibit the limbic system, an inner state emotional processing network. Furthermore, the occurrence of in-the-moment re-experiencing symptoms in patients with PTSD has been linked to increased insula activity and decreased rostral anterior cingulate cortex (rACC) and inferior frontal cortex activity [5]. Taken together, these findings suggest that PTSD involves a failure of top-down cortical inhibition (originating from the regions such as the mPFC or rACC) over the reactivation of memory traces associated with trauma related thoughts and feelings. However, studies focusing on PTSD-related neural circuits have not yet to elucidate what cause the long-term alteration in the activity pattern of engram neurons within specific brain regions implicated in PTSD.

Theoretically, cellular events in response to stress should contribute to this modulation process, which exerts a long-term effect on the synaptic strength of engram neurons, lasting from days to months. In previous studies, autophagy, a self-eating catabolic pathway in response to stress, has been shown to be altered in rodent models of PTSD [6, 7]. Presently, the autophagic process is known to be mediated by a series of autophagy-related (Atg) proteins, including Beclin-1 and microtubule-associated protein 1 light chain 3 (LC3) [8]. Beclin-1 is essential for recruitment of other Atg proteins during the early stages of autophagy [9]. Membrane anchoring of LC3-II, formed through the combination of cytosolic LC3-I with phosphatidylethanolamine, is necessary for autophagosome formation, and LC3-II levels closely correlate with the number of autophagosomes [9]. Previously, functional inhibition of Beclin-1 has been shown to play a protective role in the pathogenesis of Alzheimer’s disease [10]. Specifically, immunofluorescence and Western blotting showed that Beclin-1 levels increased from day 4 to day 7 after stress exposure, followed by a slight decline by day 14 in the mPFC of PTSD rats. This pattern was consistent with the expression of LC3-II [7]. The upregulation of autophagy may rely on inhibition of the mTOR pathway, which can be triggered by increased glucocorticoid release, aberrant neural transmission, or elevated reactive oxygen species (ROS). Previous studies have shown that enhancing autophagy may prevent the progression of a number of diseases, including Wilson’s disease, acute liver injury and acute kidney injury [11, 12]. However, this local protection may have an adverse effect on the integrated neural network of the brain. Unfortunately, much remains unknown about how autophagy affects the pathogenesis of PTSD within PTSD-related neural circuits.

In our study, we first employed single-prolong stress (SPS) as a PTSD model and confirmed the upregulation of autophagy in multiple brain regions of SPS mice in response to stress. Subsequently, autophagy was inhibited by intracerebroventricular administration of 3-Methyladenine (3-MA), a macroautophagic inhibitor. Following 3-MA treatment, anxiety-like behaviors in SPS mice were significantly alleviated. Furthermore, anxiety-like behaviors in SPS mice were alleviated by intracerebroventricular administration of lentivirus-mediated shRNA targeting expression of Atg7 (autophagy-related protein 7), which regulates autophagy. Next, we used the fear-condition (FC) task as another PTSD model, and intracerebroventricular administration of shRNA-Atg7 also facilitated fear extinction in model mice, along with reduced anxiety. Interestingly, microinjection of shRNA-Atg7 into the amygdala, rather than mPFC or hippocampus, had a similar effect on fear extinction. Additionally, Fmr1 knockout (Fmr1 KO) mice, which were previously reported to have disrupted autophagy, were investigated as a potential innate PTSD-resistant mouse model. As expected, Fmr1 KO mice exhibited fewer anxiety-like behaviors following SPS exposure. Finally, we found that local autophagy activation in amygdala reserved the anxiolytic activity in Fmr1 KO mice, which further consolidate our main finding that the upregulation of autophagy in amygdala is essential for PTSD-related anxiety.

Materials and methods

Animals

Male Wildtype mice (WT, C57BL/6) and Fmr1 KO mice (B6. 129P2-Fmr1tm1Cgr/J, the Jackson Laboratory) were purchased and bred in the animal facility at Wenzhou Medical University. WT and Fmr1 KO mice were 6-8 weeks old at the start of the experiments, unless otherwise specified. A total of 180 WT and 30 Fmr1 KO mice were randomly grouped into NT (No Treatment) or PTSD model and were switched to single housing one week prior to modeling. Mice were maintained on a 12 h:12 h light: dark cycle (lights on at 07:00), at a temperature of 20-22 °C and humidity of 30-60%, with ad libitum access to food and water. On the day of modeling, all mice were placed in the same room. Mice in the PTSD group were treated as described below, while those in NT group mice were not exposed to any stressors. Fmr1 KO mice were genotyped by polymerase chain reaction using the following primers: Fmr1 mutant forward, 5’-CAC GAG ACT AGT GAG ACG TG-3’, Wild type forward, 5’-TGT GAT AGA ATA TGC AGC ATG TGA-3’, reverse, 5’-CTT CTG GCA CCT CCA GCT T-3’. All animal experiments were conducted with the approval of the Animal Care and Use Committee of Wenzhou Medical University in Wenzhou.

pSLenti-U6-shRNA-CMV-EGFP-F2A-Puro lentivirus

Used lentivirus with the following sequence:

Raptor shRNA:

CCGGGCCCGAGTCTGTGAATGTAATCTCGAGATTACATTCACAGACTCGGGCTTTTTG

Atg7 shRNA:

GATCCCCGCAGCTCATTGATAACCATTTCAAGAGAATGGTTATCAATGAGCTGCTTTTTC

Control (Crl) shRNA:

ATCTCGCTTGGGCGAGAGT

Single-prolonged stress model

In our study, mice were exposed to single-prolonged stress (SPS), a widely used model for PTSD. This procedure is designed to induce persistent PTSD-like symptoms, and includes several severe, systematically different stressors as follows: (1) mice were restrained in 50 ml centrifuge tubes with multiple holes drilled around the tube (air holes were spaced approximately 1 cm apart) for 2 h; (2) After restraint, mice were immediately subjected to a 20-min forced swim in a 2 L beaker (diameter = 14.5 cm) containing 1.2 L water at a depth of 11.7 cm (1.2 L) and a temperature of ~23 °C; (3) Mice were then towel-dried and exposed to ether until they lose consciousness (ether exposure was performed by placing ether-soaked cotton balls in a standard microisolator polycarbonate cage); (4) Finally all mice were housed in new cages with fresh bedding and left undisturbed for seven days to allow PTSD symptoms to develop [13]. The SPS procedure was done between 9 AM and 2 PM. Control mice were not exposed to any stressors.

Foot shock stress (fear memory) paradigm

Experimental procedures were adapted from a previous study with modifications [14]. Generally, in the foot shock paradigm, mice were introduced into fear conditioning chambers (35 cm × 20 cm × 20 cm, Jiliang Tech) and allowed a 5-min adaptation period. A total of 11 intermittent, inescapable electric foot shocks (0.5 mA, 2 s each), each followed by 30 s of 80 dB white noise, were administered. Control groups mice were placed in the same conditioning chambers for an equivalent amount of time but were not subjected to foot shocks or noise. The chambers were wiped clean with 75% ethanol solution between sessions. Three days after foot shock exposure, fear-conditioned mice were reintroduced to the same chambers once daily for 10 days of extinction training. Mice in the control group were also placed in the same chambers for an equivalent amount of time each day. Spontaneous activity was recorded for two and a half minutes during extinction training session. The freezing behavior of mice was measured as an indicator of contextual and auditory fear memory induced by aversive experience. The percentage of cumulative freezing time during spontaneous activity was used to assess the fear responses of mice. Memory recall and extinction were assessed by measuring the freezing time of the first day and the last day of the extinction training sessions, as described in previous studies [14,15,16]. All mice were tested throughout the procedure.

Mouse behavioral tests

Open field test (OFT)

The OFT test was conducted in dim light to assay anxiety-related behavior [17]. Briefly, the mice were placed in a chamber (40 × 40 × 25 cm size), and they were monitored for 5 min of free movement.

In analysis, the center area of the chamber was defined as the 25×25 cm zone in center. The total distance and time spent in the central area were measured using the DigBehv software (Shanghai Jiliang Software Technology Co., Ltd.).

Elevated plus maze (EPM)

The EPM test was also used in dim light to evaluate anxiety-related behavior [18]. The cross maze is connected by two open arms (35 cm×5 cm) and two closed arms (35 cm×5 cm) and the central area (5 cm×5 cm), with a height of 45 cm. When the test started, mice were released from the center and allowed to explore the maze for 5 min. Time spent in open arms, number of open-arm entries were analyzed using the DigBehv software (Shanghai Jiliang Software Technology Co., Ltd.).

Forced swim test (FST)

The FST was employed to examine the depressive-like behavior of the mice [19]. In our study, mice were gently released into in a 2 L beaker (diameter = 14.5 cm) with a water depth of about 11.7 cm (1.2 L) at room temperature (about 23 °C). Total immobility time were analyzed by using the DigBehv software (Shanghai Jiliang Software Technology Co., Ltd.). And the stationary time was also timed manually to exclude errors.

Tail suspension test (TST)

The TST is a model of behavioral despair which can reflect depression-like behavior [20]. In the test, mice were suspended by their tail (approximately 1.5 cm from the tip of the tail) with the short-term and inescapable stress. Behavioral tests were always performed between 9AM and 6PM with minimal background noise, immobility time was recorded during a 5-min period, and all the tests were recorded by using the DigBehv software (Shanghai Jiliang Software Technology Co., Ltd.). And the stationary time was also timed manually to exclude errors.

Sucrose preference test (SPT)

The sucrose preference test can also be designed to detect the depression-like behavior. In this test, mice were placed individually in cages for 48 h, and two bottles (one bottle filled with water and one bottle filled with 1% sucrose) are placed during the adaptation phase. Fasting for food and water for 24 h after adaptation. Water and sucrose consumption were then measured at 12 h, 24 h, 48 h, 72 h. Sucrose preference index = sugar water consumption / (sugar water consumption + pure water consumption).

Western blot

The protein from mice brain (mPFC, Hippocampus and Amygdala) were extracted by using RIPA buffer containing protease inhibitor cocktail (Beyotime). Lysates were centrifuged at 12,000 rpm for 30 min under the temperature of 4 °C, and then added 5 x loading buffer and denatured at 100 °C for 10 min. Denatured protein was separated by SDS-PAGE (12%) and transferred into polyvinylidene fluoride membranes (ISEQ10100, Millipore). Membranes were blocked with blocking buffer (EpiZyme, Shanghai, China), incubated with primary antibodies at 4 °C overnight. And then the membranes were incubated with secondary antibodies at room temperature for 1 h. Finally, protein bands were developed using ECL (Bio-Rad), and image analysis was conducted using Image J. Data was normalized to GAPDH. The primary antibodies used in this study were as follows: mouse anti-GAPDH (#60004-1-Ig, Proteintech, 1:50,000), rabbit anti-Atg7 (10088-2-AP, Proteintech, 1:1,500), rabbit anti-Beclin-1 (11306-1-AP, Proteintech, 1:1,000), rabbit anti- LC3B (L7543, Sigma, 1:1,000). The secondary antibodies used in this study were as follows: goat anti-mouse IgG-HRP (#SA00001-1, Proteintech, 1:5,000) or goat anti-rabbit IgG-HRP (#SA00001-2, Proteintech, 1:5,000).

Immunofluorescence

Brain tissues were firstly fixed with 4% PFA for 72 h, and then were dehydrated in a 30% sucrose solution for another 72 h, finally they were embedded with Tissue-Tek OCT compound. And then according to the mouse brain atlas, sagittal and coronal sections were cut with a thickness of 20 μm. More than three parallel sections per mouse were used for analyses. For staining of the brain tissue sections, the tissues firstly fixed with 4% PFA for 30 min and then were undertook antigen retrieval for 30 min at 90 °C by sodium citrate antigen retrieval solution (C1032, Solarbio). After antigen retrieval, tissue sections were blocked and permeabilized with 5% BSA (4240GR100, Biofroxx) plus 0.3% Triton X-100 (T8200, Solarbio) for 1 h at room temperature and incubated overnight with primary antibodies at 4 °C. The next day, the tissues were incubated with corresponding secondary antibodies at room temperature for 1 h and finally sealed with DAPI (Solarbio). The primary antibodies used in this study were as follows: rabbit anti-Atg7 (10088-2-AP, Proteintech, 1:200), rabbit anti-Beclin-1 (11306-1-AP, Proteintech, 1:500), mouse anti-c-Fos (66590-1-Ig, Proteintech, 1:200). Secondary antibodies included Donkey anti-rabbit Alexa Fluor488 (A21206, Invitrogen, 1:1,000), Donkey anti-mouse Alexa Fluor488 (A21202, Invitrogen, 1:1,000), Donkey anti-rabbit Alexa Fluor555 (A32732, Invitrogen, 1:1,000), Donkey anti-mouse Alexa Fluor555 (A32773, Invitrogen, 1:1,000). Images were acquired using confocal microscopes (TCS SP8, Leica) or microscope (Li2, Nikon) and analyzed with Image J and Photoshop (Adobe).

Tyramide signal amplification (TSA) multicolor immunofluorescence

For immunofluorescence of mice brain sections, 5um thick sections are baked at 65 °C for 1 h. Then sections are carried out in xylene I-II for 15 min, 100% ethanol I and II for 3 min twice, 95% ethanol for 2 min, followed by 1 min each in 90%, 80%, and 70% ethanol for dewaxing. Antigen retrieval is performed using microwave repair for 5 min or high-pressure repair for 1 min with sodium citrate antigen retrieval solution (C1032, Solarbio, PH 6.0). After washing in PBS and treatment with 3% H2O2, a blocking solution is added and incubated for 30 min at room temperature. Then incubated overnight with primary antibodies at 4 °C. The next day, HRP-conjugated secondary antibodies corresponding to the primary antibody species are added and incubated at 37 °C for 30 min followed by additional PBS washes. Signal amplification is achieved through TSA dilution ranging from 1:100 to 1:1000 and developing for 10 min. The antibody complex is carefully removed using a sodium citrate buffer or a specialized removal buffer/kit. This process is iterated until all labeling is completed, culminating in a final PBS wash and sealed with DAPI (Solarbio). The primary antibodies used in this study were as follows: rabbit anti-Atg7 (10088-2-AP, Proteintech, 1:200), rabbit anti-Beclin-1 (11306-1-AP, Proteintech, 1:500), mouse anti-c-Fos (ab208942, Abcam, 1:200), rabbit anti- LC3B (L7543, Sigma, 1:200).

Stereotaxic injection surgery

3-Methyladenine(3-MA)

Male 2-month-old wild-type mice were randomly divided into four groups and then were anesthetized with tribromoethanol (250 mg/kg of body weight, i.p,). After the mouse enters the anesthesia state, the hair on the top of the head was shaved off and fixed to the stereotaxic apparatus. Then cut along the midline, carefully peel off the fascia and rub the surface of the skull with a sterile medical cotton swab dipped in a small amount of hydrogen peroxide until fully exposing the Bregma (anterior fontanelle) point. According to the “Mouse Brain Stereotaxic Atlas”, the ICV localization coordinates are selected as anteroposterior (AP) −0.2 mm, mediolateral (ML) + 1 mm (single side, right drilling hole), and dorsoventral (DV) −2.5 mm (the depth of the needle is based on the dura mater). Embed the cannula to 2.5 mm subdural at the selected coordinates. Finally, use dental tray water and dental tray powder to tightly bond the cannula, screws, skull, solidify and loosen the device that holds the cannula, gently remove the PE tube, and insert it into the core of the cannula. After the mice experienced the SPS, 3-MA was given to the ICV on day3-6 through the cannula (with a speed of 2ul/5 min). For the 3-MA groups, 3.7 ug 3-Methyladenine (HY-19312, MCE) dissolved in 0.5 μl saline (ultrasound assisted dissolution) were infused over 5 min and the injector was kept for another 5 min. For other groups, an equal volume of saline was injected.

pSLenti-shRNA-CMV-EGFP

Intracerebroventricular (ICV) injections: Unilateral intracranial injection of CMVs (OBiO) was performed stereotactically at coordinates posterior 0.2 mm, lateral 1.0 mm, and ventral 2.5 mm relative to the bregma in 2-month-old wild-type mice. 3ul of viral suspension containing 1 × 109 vector genomes (vg) were injected using a 5-μl glass Hamilton syringe with a fixed needle.

Amygdala injections

Bilateral intracranial injection of CMVs (OBiO) was performed stereotactically at coordinates posterior 1.4 mm, lateral 3.2 mm, and ventral 5.0 mm relative to the bregma in 2-month-old wild-type mice. 1ul per side of viral suspension containing 1 × 109 vector genomes (vg) were injected using a 5-μl glass Hamilton syringe with a fixed needle.

mPFC injections

Bilateral intracranial injection of CMVs (OBiO) was performed stereotactically at coordinates anterior 2.0 mm, lateral 0.4 mm, and ventral 2.4 mm relative to the bregma in 2-month-old wild-type mice. 1ul per side of viral suspension containing 1 × 109 vector genomes (vg) were injected using a 5-μl glass Hamilton syringe with a fixed needle.

HIP injections

Bilateral intracranial injection of CMVs (OBiO) was performed stereotactically at coordinates posterior 1.7 mm, lateral 1.8 mm, and ventral 2.0 mm relative to the bregma in 2-month-old wild-type mice. 1ul per side of viral suspension containing 1 × 109 vector genomes (vg) were injected using a 5-μl glass Hamilton syringe with a fixed needle.

RNA extraction and quantitative real-time PCR (qRT–PCR)

Cell RNA was extracted with RNAiso Plus (TRIzol) (T9108, Takara). The concentration of the RNA was determined by UV spectroscopy, then reverse transcribed using TransScript® All-in-One First-Strand cDNA. TransScript® All-in-One First-Strand cDNA Synthesis SuperMix (One-Step gDNA Removal) kit (AT341, TransGen Biotech) was used to reverse transcript 1 g of extracted RNA using UV spectroscopy. With the Real-Time PCR Detection System (CFX96, Bio-Rad), we quantified the expression levels of Atg7 mRNA using TransStart Top Green Supermax (AQ131, TransGen Biotech). The primers used were be listed. For Atg7 primer (Tm = 85.7 °C), forward, 5′-GTTCGCCCTTTAATAGTGC-3′; for GAPDH primer (Tm = 86.9 °C), forward, 5′-ACCCTTAAGAGGGATGCTGC-3′ and reverse, 5′-CCCAATACGGCCAAATCCGT-3′. A 20-vl reaction volume was heated at 94 °C for 30 s, followed by 40 cycles of 94 °C for 5 s and 60 °C for 30 s. Multiple changes in gene expression were calculated using the dual-Ct method, using GAPDH as an internal comparator.

Samples or animals exclusion criteria

Animals

  1. 1.

    Missing Data: If certain animals have incomplete or erroneous data during data collection, they may be excluded to ensure the accuracy of analysis.

  2. 2.

    Inconsistent Behavior: Animals that display abnormal behaviors (such as extreme anxiety or aggression) during pre-experimental behavioral assessments may be excluded to ensure uniformity in results.

  3. 3.

    Poor Adaptability: Animals that do not adapt well to the new environment during acclimatization tests may be excluded.

Samples

  1. 1.

    Insufficient Reproducibility: If the Western Blot experiments do not yield consistent results across multiple independent repeats, the data may be excluded. Insufficient reproducibility may arise from inconsistent experimental conditions or improper handling.

  2. 2.

    Technical Issues: This includes improper electrophoresis, uneven transfer, issues during the development process, etc., which may compromise the reliability of the data.

  3. 3.

    Bias in Data Analysis: During quantitative analysis, high background signals, unclear band separation, or other image processing issues may lead to inaccurate data.

Statistical analysis

“Statistical analyses were conducted using GraphPad Prism version 8.0. The Shapiro-Wilk test was used to assess the normality of the data, and the F-test was used to evaluate variance homogeneity. As the data followed a normal distribution, the student’s t-test was used for comparisons between two groups. One-way ANOVA (for single-factor analysis) and two-way ANOVA (for two-factor analysis) were performed with Tukey’s post hoc test applied for multiple group comparisons.

Results

Autophagy was upregulated in the brain of PTSD model mice

Previous studies have confirmed that SPS is an effective and reproducible animal model for PTSD [21]. C57BL/6 strain mice were subjected to SPS modeling in strict accordance with standard procedures (Fig. 1A). After modeling, we assessed anxiety-related behaviors in the mice using OFT and EPM. The results showed that mice experienced SPS spent less time in the center zone than NT mice in OFT (Fig. 1B & C), and exhibited reduced open arm entries and time in the open arms (%) in the EPM compared with NT mice (Fig. 1D & E, Fig. S1A).To rule out the possibility that stress-induced depression might confound the interpretation of molecular mechanisms underlying PTSD, we evaluated depressive-like behavior in SPS mice using the TST and FST. The results showed that SPS mice did not exhibit significant depressive-like behavior compared with NT group (Fig. S1B-E). Therefore, we aimed to determine whether the brain is the most sensitive organ in response to stress, based on the evaluation of autophagy flux. The expression level of autophagy-related proteins including Beclin-1, LC3-I/LC3-II, Atg7 were detected by Western blot in multiple organs, specifically the brain, heart and liver of SPS mice and control mice. As expected, we found that the expression of Atg7, Beclin-1 and LC3-II in the prefrontal cortex, hippocampus and amygdala was increased in the SPS group compared with NT group (Fig. 1F & H, Fig. S1F-I). Immunofluorescence staining of Beclin-1, Atg7 and LC3 in prefrontal cortex, hippocampus and amygdala also supported these findings (Fig. 1I−L, Fig. S1J, K & L). Interestingly, we also found that autophagy was only significantly upregulated in the liver of SPS mice compared with other peripheral organs (Fig. 1G & H).

Fig. 1: Autophagy was upregulated in the brain of PTSD model mice.
figure 1

A SPS procedure timeline. B OFT behavioral trajectories. C OFT center time and total distance (Student’s, t-test). D EPM behavioral trajectories. E EPM open arm entries and time (Student’s, t-test). F Western blot of Atg7, Beclin-1 and LC3 in mPFC. G Western blot of Atg7, Beclin-1and LC3 in Liver. H Quantification of Atg7, Beclin-1, and LC3-II levels in multiple organs (n = 4-5/group, One-way ANOVA, Tukey’s test). I Immunofluorescence co-staining of c-FOS, Beclin-1, Atg7, and LC3 in HIP and AMY (scale bar = 20μm). J Quantified Beclin-1 and Atg7 intensity of CA3 (Student’s, t-test). K Quantified Beclin-1 and Atg7 intensity of BLA (Student’s, t-test). L LC3 puncta count of CA3 and BLA (Student’s, t-test). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. Group sizes: NT (n = 8), SPS (n = 8); For H, J-L: n = 3 mice/group, 3 sections/mouse. See also Fig. S1.

Full size image

To further investigate the correlation between autophagy and neuronal activity after stress exposure, co-immunostaining of c-Fos and Beclin-1 was performed in SPS model mice and control mice. As a result, we found that the fluorescent intensity of Beclin-1 was positively correlated with that of c-Fos in the amygdala of SPS mice (Fig. S1N & O). These results further suggested that neurons activation is associated with the upregulation of autophagy following SPS.

Chemical inhibition of autophagy in wild-type mice after SPS can alleviate anxiety-like behavior

Our results confirmed the upregulation of autophagy in mice with PTSD. We further speculate that the anxiety-related behaviors induced by SPS are related to elevated autophagy levels. Thus, Intracerebroventricular injection of 3-methyladenine (3-MA), a known autophagy inhibitor, was administrated to WT mice after SPS modeling (Fig. 2A). Subsequently, the OFT and EPM tests were performed to evaluate anxiety-like symptoms in SPS mice with or without 3-MA treatment. We found that SPS mice treated with 3-MA spent more time in the central area of the open field arena in the OFT and exhibited more open arm entries in EPM rather than SPS mice alone, indicating that 3-MA treatment alleviated anxiety-like behaviors induced by stress (Fig. 2B−E). To assess changes in autophagy levels, we examined the expression of the autophagy-related proteins Beclin-1 and LC3 in the prefrontal cortex, hippocampus and amygdala by western blot (Fig. 2F−H, Fig. S2A−D). The results suggest that autophagy levels were significantly inhibited following 3-MA administration. Taken together, our results indicate that chemical inhibition of autophagy in SPS mice can alleviate anxiety-like behaviors induced by stress.

Fig. 2: 3-methyladenine as an Autophagy Inhibitor Can Alleviate Anxiety-like Behavior.
figure 2

A Experimental timeline of 3-MA stereotactic injection in SPS model. B OFT behavioral trajectories. C OFT center time and total distance. D EPM behavioral trajectories. E EPM open arm entries and time. F Western blot of Beclin-1 and LC3 in mPFC. G Quantified Beclin-1 level (n = 3/group). H Quantified LC3-II level (n = 3/group). Data are presented as mean ± SEM. One-way ANOVA with Tukey’s test was used for statistical analysis. *p < 0.05, **p < 0.01, ***p < 0.001. Group sizes: NT + NS (n = 8), NT + 3-MA (n = 8), SPS + NS (n = 8), SPS + 3-MA (n = 8). See also Figs. S1 and S2.

Full size image

Genetic inhibition of autophagy in wild-type mice after SPS can alleviate anxiety-like behavior

To further confirm the effect of autophagy on anxiety-like behavior in PTSD model mice, autophagy in SPS mice was genetically inhibited by intracerebroventricular administration of lentivirus-mediated shRNA targeting the expression of Atg7, a key regulator of autophagy (Fig. 3A). The knock-down efficiency of shRNA-Atg7 was confirmed by immunohistochemistry, western blot and qPCR (Fig. 3F−J, Fig. S2E−H). As a result, SPS mice treated with shRNA-Atg7 showed more open arm entries in the EPM compared with those treated with control shRNA, although there was no significant difference in time spent in the central area of the open field arena in the OFT (Fig. 3B-E). To further confirm the effect of shRNA-Atg7 on the PTSD related phenotype, WT mice were fear conditioned, which is also widely accepted as a PTSD model (Fig. 4A−D). As expected, conditioned mice injected with shRNA-Atg7 displayed more open arm entries in EPM test (Fig. 4E−H). In addition, these mice displayed enhanced memory extinction without impairing memory recall, as indicated by the freezing time during the extinction training session (Fig. 4C, D). Together, our results show that genetic inhibition of autophagy can partially alleviate PTSD-like behavior phenotypes induced by stress.

Fig. 3: Knockdown of Atg7 in Lateral Ventricle Can Alleviate Anxiety-like Behavior Caused by SPS Model.
figure 3

A Experimental timeline of shCrl/shAtg7 stereotactic injection in SPS model. B Open field test trajectories. C Center time and total distance in OFT (One-way ANOVA, Tukey’s test). D Elevated plus maze trajectories. E Open arm entries and time in EPM (One-way ANOVA, Tukey’s test). F Atg7 immunofluorescence in hippocampus (scale bar=20μm). G Quantified Atg7 fluorescence intensity (n = 3 mice/group, 3 sections/mouse). H Western blot of Atg7, Beclin-1, LC3 in HIP. I Quantified Atg7, Beclin-1, LC3-II levels (n = 3/group). J qPCR in HT22 cells (Student’s, t-test). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. Group sizes: shCrl+NT (n = 9), shAtg7+NT (n = 10), shCrl+SPS (n = 9), shAtg7+SPS (n = 10). See also Fig. S2.

Full size image
Fig. 4: Knockdown of Atg7 in Lateral Ventricle Can Also Alleviate Anxiety-like Behavior Caused by Foot shock Model.
figure 4

A Timeline and schematic of shCrl/shAtg7 stereotactic injection in foot shock model. B Behavioral paradigm of foot shock and fear memory extinction training. C Freezing time during extinction training (Two-way ANOVA, Sidak’s test). D Freezing time during recall and extinction section (Student’s, t-test). E Representative OFT trajectories. F OFT center time and total distance (One-way ANOVA, Tukey’s test). G Representative EPM trajectories. H EPM open arm entries and time (One-way ANOVA, Tukey’s test). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. Group sizes: shCrl+NT (n = 6), shAtg7+NT (n = 6), shCrl+FS (n = 6), shAtg7+FS (n = 6). See also Fig. S2.

Full size image

Local knockdown of Atg7 in amygdala rather than prefrontal cortex or hippocampus in wild-type mice after SPS can alleviate anxiety-like behavior

Given the distinct roles and neuronal activity patterns of the amygdala in the pathogenesis of PTSD, we speculate that local inhibition of autophagy in the amygdala may influence anxiety-like behavior in PTSD model mice. Bilateral injection of shRNA-Atg7 into the amygdala were performed before SPS modeling, and the OFT and EPM test were conducted to evaluate anxiety-like behavior (Fig. 5A & Fig. S2K). As a result, local injection of shRNA-Atg7 into the amygdala alleviated anxiety-like behaviors induced by SPS modeling (Fig. 5A−E). Although the amygdala plays a critical role in anxiety-like behavior, other brain subregions such as mPFC and hippocampus are also involved in the pathogenesis of PTSD. Then, bilateral injection of shRNA-Atg7 into PFC or hippocampus was performed in WT mice before modeling (Fig. S2L & M). Unexpectedly, neither autophagy inhibition in the PFC nor in the hippocampus had any effect on anxiety-like behaviors induced by stress (Fig. 6A−I). In addition, mice treated with shRNA-Atg7 in the amygdala exhibited no change in memory extinction after undergoing fear conditioning (Fig. S2I & J). Our results suggest that autophagy, which is widely activated in multiple brain subregions in response to stress, may not exert uniform effects on the pathogenesis of PTSD. Local inhibition of autophagy in amygdala may be sufficient to alleviate anxiety-like behavior in PTSD model mice.

Fig. 5: Local Knockdown of Atg7 in Amygdala Can Alleviate Anxiety-like Behavior.
figure 5

A Timeline and injection schematic in AMY with SPS model. B Open field test trajectories. C Center time and total distance in OFT. D Elevated plus maze trajectories. E Open arm entries and time in EPM. Data are presentesd as mean ± SEM. One-way ANOVA with Tukey’s test was used for statistical analysis. *p < 0.05, **p < 0.01, ***p < 0.001. Group sizes: shCrl+NT (n = 9), shAtg7+NT (n = 9), shCrl+SPS (n = 8), shAtg7+SPS (n = 9).

Full size image
Fig. 6: Local Knockdown of Atg7 in Prefrontal Cortex or Hippocampus Can’t Alleviate Anxiety-like Behavior.
figure 6

A Timeline and injection schematic in mPFC and HIP with SPS model. B Open field test trajectories with stereotaxic injection group in mPFC. C Center time and total distance in OFT. D Elevated plus maze trajectories with stereotaxic injection group in mPFC. E Open arm entries and time in EPM. F Open field test trajectories with stereotaxic injection group in HIP. G Center time and total distance in OFT. H Elevated plus maze trajectories with stereotaxic injection group in HIP. I Open arm entries and time in EPM. Data are presented as mean ± SEM. One-way ANOVA with Tukey’s test was used for statistical analysis. *p < 0.05, **p < 0.01, ***p < 0.001. Group sizes (B-E): shCrl+NT (n = 6), shAtg7+NT (n = 6), shCrl+SPS (n = 6), shAtg7+SPS (n = 6). Group sizes (F-I): shCrl+NT (n = 7), shAtg7+NT (n = 7), shCrl+SPS (n = 6), shAtg7+SPS (n = 7).

Full size image

Less pronounced anxiety-like behavior exhibited in Fmr1 KO mice can be reversed by local activation of autophagy in amygdala

A previous study has shown that the autophagic flux is significantly downregulated in hippocampal neurons in Fmr1 KO mice [22]. Therefore, we wondered whether Fmr1 KO mice could be an innate PTSD-resistant model. We first confirmed that autophagic activity in the prefrontal cortex and hippocampus of Fmr1 KO mice was downregulated, as indicated by decreased levels of Beclin-1 and LC3-II protein level (Fig. 7A−D). Then male Fmr1 KO mice were subjected to SPS protocol, and anxiety-like behaviors were evaluated using the OFT and EPM. The results showed that Fmr1 KO mice exhibited relatively less anxiety than wild-type mice after experiencing SPS after experiencing SPS, as reflected by more time spent in the center zone in the OFT and a higher frequency of entries into the open arms in the EPM (Fig. 7E & F). Thus, our results suggest that individual differences in autophagy activity may partially contribute to susceptibility to PTSD.

Fig. 7: Less pronounced anxiety-like behavior exhibited in Fmr1 KO mice can be reversed by local activation of autophagy in amygdala.
figure 7

A Western blot analysis of LC3 and Beclin-1 levels in mPFC. B Quantification of LC3-II and Beclin-1 levels in mPFC (n = 3/group). C Western blot analysis of LC3 and Beclin-1 levels in HIP. D Quantification of LC3-II and Beclin-1 levels in HIP (n = 3/group). E Center time and total distance in OFT. F Open arm entries and time in EPM. G Timeline and stereotactic injection schematic for shCrl/shRaptor in AMY. H Center time and total distance in OFT. I Open arm entries and time in EPM. Data are presented as mean ± SEM. One-way ANOVA with Tukey’s test was used for statistical analysis. *p < 0.05, **p < 0.01, ***p < 0.001. Group sizes: WT + NT (n = 7), Fmr1 KO + NT (n = 7), WT + SPS (n = 7), Fmr1 KO + SPS (n = 7). Group sizes: WT+shCrl (n = 8), Fmr1 KO+shCrl (n = 8), WT+shRaptor (n = 8), Fmr1 KO+shRaptor (n = 7).

Full size image

As demonstrated above, local inhibition of autophagy in amygdala may sufficiently alleviate anxiety-like behavior in PTSD model mice. To further confirm our major finding, bilateral injection of Raptor shRNA virus vector int o the amygdala was applicated in Fmr1 KO mice. Raptor shRNA has previously been confirmed to active autophagy [22]. As expected, the anxiolytic effect in Fmr1 KO mice was reversed by local autophagy activation, as reflected by less time spent in the center zone in the OFT and a reduced frequency of entries into the open arms in the EPM in shRaptor-treated Fmr1 KO mice (Fig. 7H & I). Taken together, our results suggest that autophagic activity in the amygdala plays a pivotal role in the pathogenesis of PTSD-related anxiety.

Discussion

Our study investigated the impact of autophagy on the pathogenesis of PTSD in the context of PTSD-related neuronal circuits. In this study, we found that both chemical and genetic inhibition of autophagy can alleviate anxiety-like behaviors and fear-related memory induced by a PTSD model. Moreover, we found that local genetic inhibition of autophagy in the amygdala, rather than in the mPFC or hippocampus can alleviate anxiety-like behaviors induced by a PTSD model. In addition, our result suggests that autophagy deficiency renders Fmr1 KO mice a PTSD-resistant model, and local activation of autophagy in the amygdala can reverse their anxiolytic phenotype.

Autophagy, a conserved intracellular process in eukaryotes, regulates the turnover of proteins and organelles. Generally, autophagy serves as a protective mechanism for various organ systems during stress. For instance, pharmacological activation of autophagy by rapamycin protects against acetaminophen-induced acute liver injury, whereas inhibition of autophagy by inducible deletion of Atg7 in hepatocytes further enhances liver damage [23, 24]. Even in neuronal system, pharmacological enhancement of autophagy can alleviate neuronal death in oxidative stress-induced Parkinson’s disease models and genetic model mice of Alzheimer’s disease [25, 26]. However, upregulation of autophagy that is protective in individual neurons may be detrimental to higher brain function and the integration of the entire neuronal system. In addition, uncontrolled autophagic activity can lead to excessive neuronal death, which has been implicated in certain psychiatric disorders.

In this study, we first employed SPS procedure to establish a PTSD mouse model. Previous studies have demonstrated that the SPS model of traumatic stress exposure shows significant validity as an animal model of PTSD, inducing symptoms resembling those observed in human PTSD, primarily increased anxiety [27]. According to the DSM-IV, PTSD is defined as a severe anxiety disorder that develops after a traumatic event. The appearance of anxiety symptoms may increase the risk of PTSD [28]. As expected, mice exhibited significant anxiety-like behaviors without depressive symptoms following SPS modeling. Subsequently, we postulated that, compared with peripheral organs, the autophagic process in the brain is particularly sensitive to traumatic stress, which has not been previously investigated. Interestingly, we found that autophagic flux in the liver, rather than in other peripheral organs, was also significantly upregulated in SPS mice. Following the above reasoning, upregulated autophagy may protect the liver from the stress-induced cell damage, however, this liver-restricted response may not be fully explained by activation of the endocrine system. Future studies could investigate whether the liver-specific impact of PTSD model depends on innervation by specific neuronal circuits.

To investigate whether anxiety-like symptoms of PTSD mouse can be attributed to excessive activation of autophagy in the brain, we first used 3-MA, an autophagy inhibitor, to suppress the level of autophagy in the relevant brain area by means of lateral ventricular injection, which could exclude the effect of peripheral tissues on behavior. Previous study suggests that hepatic response to stress enhances depressive-like behaviors by releasing 14,15-epoxyeicosatrienoic acid [29]. In light of autophagy was upregulated in liver in our study, we employed this mode of administration. As expected, anxiety-like symptom was relieved. Further, we genetically inhibit autophagy by intracerebroventricular administration of lentivirus shRNA-Atg7 vector, however, only performance in EPM test was rescued. We think that this could potentially be attributable to the diffusion velocity and ranger of drug dissemination, rather than virus administration, which in turn make the difference in brain regions affected by the inhibition of autophagy. As we said, we were most interested whether the impact of autophagy on PTSD is brain-region specific. The over-activation of amygdala neurons has been widely believed to contribute to core phenotypes of PTSD, whereas the involvement of mPFC in PTSD, may be more complicated. In some case, disrupted neuronal activity in mPFC in PTSD was observed, however, in other case enhanced neuronal activity in mPFC causes the activation of amygdala through direct projection [30,31,32]. It is hard to believe that autophagy has distinct effects on neurons in amygdala or mPFC. As we expected, only the inhibition of autophagy in amygdala had an effect on anxiety-like behavior, showing significant differences in EPM but not OFT. Compared with EPM test, OFT performance may reflect subtle but distinct behavior characteristic of anxiety, which may not only depend on the activation of amygdala [33]. Another noteworthy point of our findings is that, in the conditioned fear memory, enhanced memory extinction observed with general autophagy inhibition in the brain was not replicated by local autophagy inhibition in the amygdala. This finding further suggests that upregulated autophagy in the amygdala, which appears to be exclusively responsible for the expression of anxiety-like behaviors, may not be involved in fear memory extinction. It also supports the notion that distinct brain subregions within the PTSD-related neuronal circuit may contribute to different symptoms [4].

To consolidate our finding that autophagic activity in the amygdala is essential for the pathogenesis of anxiety-like behaviors in a PTSD mouse model, local activation of autophagy in an innate PTSD-resistant mouse model might be a promising strategy. To date, human genomic studies have revealed no obvious genetic associations between autophagy related genetic variability and the risk of PTSD onset, which could be interpreted as suggesting that autophagy deficiency may result in much more severe disease rather than psychiatric problem. We still questioned whether the onset of PTSD could be innately affected by autophagy. One previous study showed that autophagy levels were downregulated in the hippocampal neurons of Fmr1 KO mice, as evidenced by assessments of autophagic flux, functional readouts, and biochemical markers [22]. Notably, prior to PTSD model induction, we found that Fmr1 KO mice exhibited less anxiety-like behaviors compared to WT mice, which is consistence with previous studies of Fmr1 KO mice [34]. Interestingly, patients with Fragile X syndrome exhibit distinct patterns of anxiety and tend to be more aggressive [35]. Subsequently, we found that Fmr1 KO mice exhibited significantly lower levels of anxiety symptom after modeling compared to wild-type mice subject to the same SPS procedure. Furthermore, we found that local autophagy activation in the amygdala brought behavior performance in OFT and EPM in Fmr1 KO mice to a level equivalent to that of WT mice, which further support our conclusion that autophagy activity in amygdala is essential for anxiety. From an evolutionary perspective, the reduced anxiety-like performance in Fmr1 KO mice may not be advantageous, as it might reduce alertness and coping responses when facing natural enemies [36]. This could be an interesting direction for investigating the long-term consequence of premature autophagy intervention on behavior.