Ketamine preserves perineuronal nets in chronic stressed mice brain through the inhibition of microglia activation
Authors: Jingwen Yang, Ting Huang, Meng Wang, Yingrui Xu, Xuan Li, Changqi Li, Wenyu Cao, Liyan He, Huamao Zhou
Published: 2026

ABSTRACT
AbstractBackgroundDeeper understanding of ketamine’s mechanism of action would contribute to the discovery of novel therapeutic targets with fast-onset actions. To gain insight on the mechanism underlying the antidepressant-like effects of ketamine, we focused on perineuronal nets (PNNs), an extracellular matrix structure that surrounds fast-spiking parvalbumin-positive interneurons and regulates synaptic plasticity, whose integrity is known to be compromised under stress-induced depressive conditions.MethodsWe first performed the description of the plastic remodeling of PNNs in the Chronic restraint stress (CRS) mice treated or un-treated with ketamine, by quantifying the number of WFA, marker of PNNs, parvalbumin (PV) and c-Fos positive cells, as a surrogate of neuronal activity. And then, we investigated the the transcripts of a number of proteins involved in the formation or degradation of PNNs with or without ketamine in CRS mice. Next, we evaluated the expression of IBA1, a microglial marker, in the hippocampus and medial prefrontal cortex after CRS treated or un-treated with ketamine.ResultsWe found that ketamine effectively alleviated the animals’ depression like behavior as well as attenuates CRS-induced reduction of WFA-positive cells in the hippocampus and medial prefrontal cortex of mice. Although ketamine treatment had little or no effect on the number of PV, c-Fos positive cells and the transcripts of proteins involved in the formation or degradation of PNNs. Notably, ketamine treatment lead to remarkably reduced the number of IBA1-positive cells in the hippocampus and medial prefrontal cortex after CRS.ConclusionOur findings suggest that PNNs is characterized by region-specific changes in chronic stress mouse brain and provide extensive evidence that ketamine exposure initiates microglia to remodel PNN, instead of the PNN accumulation and degradation enzymes.Keywords: Depression, Ketamine, Perineuronal nets, Stress, Microglia

Introduction

Prolonged or repeated exposure to physical and psychological stress is a major risk factor for emotional disorders, including major depressive disorder (MDD) (Deng et al., 2016, Li et al., 2025, Daviu et al., 2019, World Health Organization, 2017). Although traditional antidepressants are widely used, they typically require weeks to months to achieve therapeutic efficacy, and approximately one-third of patients fail to respond adequately to treatment (Trivedi et al., 2006). A landmark study by Berman et al. demonstrated that a single subanaesthetic dose of intravenous ketamine (0.5 mg/kg infused for 40 min) exerted rapid antidepressant effects in patients with depression (Berman et al., 2000). Since then, accumulating evidence confirmed that a single low dose of ketamine induces rapid antidepressant effects within hours that can persist for approximately one week (Diazgranados et al., 2010, Fukumoto and Chaki, 2015, Rawat et al., 2022). Given ketamine’s short elimination half-life and its clearance from the body within 24 h (Maxwell et al., 2006, Zhao et al., 2012), these sustained antidepressant effects are particularly remarkable and have generated considerable interest in the underlying neurobiological mechanisms.

Ketamine was initially considered as a non-competitive N-methyl-D-aspartate receptor (NMDAR) antagonist and has long been used as an anaesthetic agent (Berman et al., 2000, Krystal et al., 2019). Activation of the ventral hippocampus (vHip)-medial prefrontal cortex (mPFC) circuit is necessary and sufficient for ketamine’s antidepressant effects (Carreno et al., 2016). However, recent studies suggest that ketamine may have multiple molecular targets (Li et al., 2010), including the widely expressed NMDAR (Ma et al., 2023). Owing to its pharmacological promiscuity, ketamine likely modulates neural activity across many brain regions and cell types, including neurons and glial cells (Davoudian et al., 2023, Stenovec et al., 2020). In addition, ketamine exerts sustained antidepressant-like effects through the facilitation of GABA function in the mPFC and hippocampus (Fogaça et al., 2025, Tornese et al., 2019), as evidenced by increased GABA levels in the mPFC of patients diagnosed with MDD following ketamine treatment (Milak et al., 2016, Rodriguez et al., 2015).

Both intracellular and extracellular proteins are involved in brain function over time (Hrabetova et al., 2018, Huntley, 2012). Many intracellular proteins exhibit relatively rapid turnover, with synthesis and degradation occurring over hours to several days (Cohen et al., 2013, Cohen and Ziv, 2017). In contrast, extracellular matrix (ECM) molecules are considerably more stable and are largely protected from intracellular ubiquitin-mediated and enzymatic degradation processes (Cohen et al., 2013, Toyama et al., 2013). Therefore, it is reasonable to speculate that the ECM may play a role in the sustained antidepressant effects of ketamine (Donegan J and Lodge D, 2017). However, compared with the well-characterised role of intracellular mechanisms, the role of ECM in mediating ketamine’s antidepressant actions remains poorly understood. A better understanding of the ECM-related mechanisms underlying this condition may contribute to the discovery of novel therapeutic targets that can produce rapid and sustained antidepressant effects.

Perineuronal nets (PNNs) are highly condensed, specialised ECM structures that predominantly enwrap parvalbumin-positive interneurons in a mesh-like manner (Kaushik et al., 2021, Lupori et al., 2023). PNNs regulate the onset and completion of critical periods during postnatal brain development, modulate neuronal excitability and synaptic transmission, and have been implicated in several brain disorders such as Alzheimer's disease, stroke, epilepsy, depression, autism, and schizophrenia (Berezin et al., 2014, Miyata and Kitagawa, 2015, Morawski et al., 2014, Riga et al., 2017, Soleman et al., 2013). The main components of PNN are hyalectans, which comprise four major families: aggrecan (ACAN), brevican (BCAN), versican (VCAN) and neurocan (NCAN) (Zimmermann and Dours-Zimmermann, 2008, Carulli et al., 2006). Wisteria floribunda agglutinin (WFA), one of the most widely used markers for visualising PNNs (Härtig et al., 2022), selectively binds N-acetylgalactosamine beta−1 (GalNAc beta−1–3 Gal) residues of glycoproteins in the PNNs (Hilbig et al., 2001).

Traditionally, PNNs have been considered physical barriers that maintain local ionic homeostasis and restrict synaptic plasticity (Fawcett et al., 2019). However, recent reports suggest a more dynamic role. Even after maturation, PNNs can undergo remodelling in response to stress or learning experiences (Sorg et al., 2016, Beurdeley et al., 2012). Enzymatic degradation of PNNs in the adult visual cortex reactivates juvenile ocular dominance plasticity (Pizzorusso et al., 2002), whereas similar manipulations in the hippocampus or mPFC impair fear conditioning and systems consolidation (Hylin et al., 2013, Shi et al., 2019). Microglia, the resident immune cells of the brain, can be primed by chronic stress (Hartmann, 2025) and have been shown to regulate PNN in various neurological disorders (Crapser et al., 2020b, Nguyen et al., 2020). The importance of PNNs in depression is further highlighted by reports demonstrating transient remodelling of mPFC PNNs following stress and ketamine treatment (Yu et al., 2022). Notably, enzymatic digestion of PNNs using chondroitinase ABC (ChABC) abolishes the sustained antidepressant effects of ketamine, indicating a functional interaction between ketamine and PNN organisation (Donegan J and Lodge D, 2017). Furthermore, regional alterations in WFA-labelled PNNs have been reported previously (Matuszko et al., 2017). Nevertheless, the precise contribution of PNN remodelling to chronic stress across different brain regions remains unclear. Moreover, despite the physiological and pathological significance of PNNs, systematic investigation of PNN alterations before and after ketamine treatment in depression models remains limited.

In the present study, we investigated the brain-wide effects of ketamine on PNN remodelling in chronic restraint-stressed (CRS) mice, a well-established model that recapitulates key molecular and physiological features of MDD (Luhong et al., 2024, Camargo et al., 2025). We first performed a WFA-based screening to identify brain regions potentially involved in ketamine-induced PNN remodelling. Based on the magnitude of the changes and the established roles of the hippocampus and prefrontal cortex in stress processing and adaptation (Yu et al., 2022, Xu et al., 2023), these regions were selected for further analysis. We further examined whether PNN remodelling was associated with alterations in PV expression or neuronal activation in a brain region-specific manner. To elucidate the molecular mechanisms underlying ketamine-induced PNN remodelling, we quantified the transcripts of proteins involved in PNN formation and degradation. In addition, we assessed microglia activation because of its established role in regulating PNN formation (Crapser et al., 2021, Tansley et al., 2022). Collectively, this study contributes to a deeper understanding of the mechanism underlying the antidepressant effects of ketamine.

Materials and methods

Animals

Eight-week-old male C57BL/6 J mice were obtained from Hunan SJA Lab Animal Co., Ltd. Animals were maintained under standard laboratory conditions (22 ± 2 ℃, 50 ± 5% humidity, and a 12-h light/dark cycle) with ad libitum access to food and water. All experimental procedures were performed in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals and were approved by the Animal Ethics Committee of the University of South China.

A total of 40 mice were used in this study, including 14 mice in the CON + NS group, 14 mice in the CRS + NS group, and 12 mice in the CRS + KET group. All behavioural tests were conducted between 08:00 and 12:00.

Chronic restraint stress

Mice were randomly assigned to one of three groups: (1) CON + NS group, where mice were housed in a polypropylene box (545 ×395 ×200 mm) and received intraperitoneal (i.p.) saline injections; (2) CRS + NS group, where mice were singly housed in polypropylene boxes (290 ×178 ×160 mm) and subjected to chronic restraint stress by placement in a ventilated restraint cylinder that severely restricted movement for 6 h daily (9:00–15:00) over 35 consecutive days (Luhong et al., 2024). Mice received i.p. saline injections; (3) CRS + KET group, in which mice underwent the same CRS procedure and received an i.p. ketamine injection (10 mg/kg).

Drugs and reagents

At the end of the CRS procedure, ketamine (KET) was dissolved in saline and administered intraperitoneally (i.p.) at a dose of 10 mg/kg (Yu et al., 2022). Behavioural testing was performed on the second day, as illustrated in Fig. 1.

Ketamine alleviates depression-like behaviours in CRS mice. (A) Experimental timeline of CRS exposure and behavioural testing. Mice were assigned to three groups: control mice receiving saline (CON+NS), CRS-exposed mice receiving saline (CRS+NS), and CRS-exposed mice receiving ketamine (10 mg/kg, intraperitoneally; CRS+KET). (B) Movement trajectories in the open-field test (OFT). (C) Total distance travelled in the OFT (n = 14, 14, and 12 mice, respectively). (D) Sucrose preference test (SPT) results (n = 14, 13, and 10 mice, respectively). (E) Immobility time in the forced swimming test (FST) (n = 14, 14, and 12 mice, respectively). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. A total of 40 mice were used, and three animals were excluded from the final analysis.

Behavioural tests

Open field test (OFT)

The OFT was performed as previously described (Xu et al., 2023) to assess anxiety and motor activity. The apparatus consisted of a square arena (50 × 50 × 50 cm), with the floor divided into 25 equal squares. Each mouse was placed in the centre of the arena and allowed to explore freely for 5 min. The total distance travelled in the central area was recorded and analysed using SuperMaze software (Shanghai XinRuan Information Technology Co., Ltd.).

Sucrose preference test (SPT)

The SPT was performed as previously described (Xu et al., 2022). Briefly, mice were individually housed and provided free access to two bottles containing either 1% sucrose solution or water. Bottle positions were switched every 12 h to eliminate side preference. The weights of both bottles were recorded at baseline and after 12 and 24 h. Sucrose preference was calculated as: Sucrose preference (%) = (sucrose consumption / (water consumption + sucrose consumption)) × 100%.

Forced swimming test (FST)

The FST was performed as previously described (Shi et al., 2023). Mice were placed in a transparent glass cylinder (25 cm height, 15 cm diameter) filled with water (24 ± 1℃) to a depth of 17 cm, preventing contact with the bottom. The test lasted 6 min, and the immobility time during the final 4 min was recorded. Immobility was defined as the absence of struggle, excluding necessary movements to keep the head above water. Data were analysed using SuperFst software (Shanghai XinRuan Information Technology Co., Ltd.).

Immunohistochemistry

Animals were euthanised with an overdose of sodium pentobarbital (10%, 80 mg/kg) and perfused with normal saline followed by 4% paraformaldehyde. Brains were collected and fixed overnight in 4% paraformaldehyde at 4℃, and subsequently cryoprotected in 15% and 30% sucrose solutions. Coronal sections (30 μm) were prepared using a freezing-sliding microtome (Leica). Sections were incubated in 3% hydrogen peroxide for 25 min and washed three times with 0.01 M PBS (10 min each). After blocking with 5% rabbit or goat serum for 2 h at room temperature, sections were incubated with primary antibodies (Table 1) for 2 h at room temperature and then overnight at 4℃. Following PBS washes, sections were incubated with biotinylated rabbit anti-goat IgG or goat anti-rabbit IgG for 2 h at room temperature for IBA−1, PV, c-fos, aggrecan, and neurocan. No secondary antibodies were required for WFA staining. After additional PBS washes, sections were incubated with avidin-biotin complex (ABC; 1:200, Vector Laboratories) for 2 h at room temperature. Subsequently, diamine-benzidine tetrahydrochloride (DAB) was used as a peroxidase substrate. Stained sections were mounted, dried, dehydrated, degreased, and sealed with neutral resin glue. Images were acquired under identical conditions using an optical microscope (EX30, Shunyu, China).

Table 1Antibodies or reagents used in this study.Antibody or reagentCompanyDilution ratiobiotinylated WFAVector Labs1:1000goat anti-IBA1Abcam1:1000rabbit anti-PVABconal1:1000rabbit anti-c-fosCST1:1000rabbit anti-Aggrecanproteintech1:2000rabbit anti-NeurocanBoster1:500biotinylated rabbit anti-goat IgGproteintech1:200biotinylated goat anti-rabbit IgGproteintech1:200ABCVector Labs1:200

Antibodies or reagents used in this study.

Real-time PCR

Animals were euthanised with an overdose of sodium pentobarbital (10%, 80 mg/kg) followed by decapitation. The hippocampus and medial prefrontal cortex were separated and stored at −80℃ until analysis. Total RNA was extracted using Trizol reagent (CWBIO) according to the manufacturer’s instructions. RNA purity was assessed by measuring the A260/A280 ratio. Complementary DNA (cDNA) was synthesised using a cDNA synthesis kit (Thermo Scientific). Quantitative real-time PCR was performed using an ABI real-time PCR system and TB Green® Premix Ex Taq™ II (TaKaRa) according to the manufacturer’s protocol. GAPDH served as the internal control. Primer sequences were designed using Primer 3 software and are listed in Table 2. The amplification protocol consisted of an initial denaturation step at 95℃ for 30 s, followed by 45 cycles of 95℃ for 10 s and 60℃ for 30 s. All samples were processed in technical duplicates, and melting-cure analysis was performed to confirm amplification specificity. Relative gene expression was calculated using the 2-△△Ct method as previously described (Zhang et al., 2025a).

Table 2Lists of primer sequences.GenesPrimersSequence (5’−3’)GAPDHForwardACCACCATGGAGAAGGCTGGGAPDHReverseCTCAGTGTAGCCCAGGATGCMMP2ForwardCAAGTTCCCCGGCGATGTCMMP2ReverseTTCTGGTCA AGGTCACCTGTCMMP9ForwardGCTGACTACGATAAGGACGCAMMP9ReverseTAGTGGTGCAGGCAGAGTAGGAADAMTS1ForwardCTCTCACCCTTCGGAATTTCTGAMAMTS1ReverseGGAGCCACATAAATCCTGTCTGAMAMTS4ForwardCTCTCACCCTTCGAATTTCTGAMAMTS4ReverseGGAGCCACATAAATCCTGTCTGAMAMTS5ForwardCTGCCTTCAAGGCAAATGTGTGAMAMTS5ReverseCAATGGCGGTAGGCAAACTGCACSGalNAcT1ForwardCTGGAGAGTCTAAACAGCCCTGCSGalNAcT1ReverseAAGGTCAGCTCATACAAAGTGCCCSGalNAcT2ForwardCAAGAGGATTCATCTCACCGTTGCSGalNAcT2ReverseGCACCCACATTTAGTCCTCGTCAggrecanForwardGGAGACCCAGACAGCAGAAACAACAggrecanReverseGCAGGTGGCTCCATTCAGACAAGNeurocanForwardGGAGTAGATATGGAAGCGACA ANeurocanReverseTGTCGACCATTCAACCCTTTAABrevicanForwardGCACCACCTACCGAGTCCTCAGBrevicanReverseGGTCCTCCTGCTCCTTCTCTTCCTenascin RForwardTGTTGCGCCTTTTGATTACTACTenascin RReverseGATGTTGGTAGCGATCAAATCCPhosphacanForwardCTGCACTTTACGTCTGTACCTAPhosphacanReverseACTCGAGTGAGGAACTGTTAAG

Lists of primer sequences.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 8.0.1. Data are presented as mean ± SEM. The assumptions of normality were assessed using the F-test or Bartlett’s test. For normally distributed data, comparisons between two groups were determined using Student’s t-test, whereas comparisons among multiple groups were analysed using either one-way ANOVA followed by Bonferroni post-hoc testing or Welch’s ANOVA followed by Games-Howell’ s multiple comparisons test. For non-normally distributed data, the Kruskal–Wallis test was used, followed by Dunn’s post hoc test with Bonferroni correction for multiple comparisons. Statistical significance was defined as P < 0.05. Data points deviated more than 2.5 standard deviations from the group mean were considered outliers and excluded from subsequent analyses.

Result

Ketamine alleviated depression-like behaviours in CRS mice

We first validated the antidepressant-like effects of ketamine in the chronic restraint stress (CRS) mouse model (Fig. 1A). No significant differences in the total distance travelled were observed among the three groups in the open-field test (F(2, 37) = 0.7127, P = 0.4969; Fig. 1B–C). In contrast, sucrose preference differed significantly among groups (H(2) = 9.931, P = 0.0070). Dunn's post hoc analysis revealed that CRS significantly decreased sucrose preference compared with the CON + NS group (P = 0.0169), whereas ketamine treatment restored sucrose preference in CRS mice (P = 0.0249).

Similarly, significant group differences were observed in the forced swimming test (FST) (F(2, 37) = 18.60, P < 0.0001; Fig. 1D–E). Bonferroni post hoc analysis showed that CRS increased immobility time relative to controls (t = 6.084, P < 0.0001, df = 15), and this increase was reversed by ketamine treatment (t = 3.304, P = 0.0064, df = 15).

Ketamine attenuated CRS-induced reductions in WFA-positive cells in the hippocampus and medial prefrontal cortex

To investigate the effects of ketamine on perineuronal nets (PNNs), we quantified WFA-positive cells in multiple brain regions using immunohistochemistry (Fig. 2A). Significant group differences in both the hippocampus (HIP) (F(2, 31) = 18.41, P < 0.0001; Fig. 2B) and medial prefrontal cortex (mPFC) (F(2, 29) = 7.021, P = 0.0033; Fig. 2I). Compared with controls, CRS markedly reduced the number of WFA-positive cells in both regions (HIP: t = 5.688, P < 0.0001, df = 31; mPFC: t = 3.394, P = 0.0060, df = 29), whereas ketamine treatment restored WFA-positive cell numbers (HIP: t = 4.873, P < 0.0001, df = 31; mPFC: t = 3.126, P = 0.0120, df = 29).

Ketamine attenuates CRS-induced reductions in WFA-positive cells in the hippocampus and medial prefrontal cortex. (A) Experimental design for WFA immunohistochemical staining across different brain regions. (B) Representative images of WFA-positive cells in the hippocampus (HIP) (Bar = 100 μm). (C) Quantification of WFA-positive cells in the HIP (n = 3 mice per group, two brain sections per mouse; left and right hemispheres quantified independently). Each data point represents measurements from one side of a single section. (D) Representative images of WFA-positive cells in hippocampal subregions (Bar = 100 μm). (E–H) Quantification of WFA-positive cells in (E) CA1, (F) CA2, (G) CA3, and (H) DG (n = 3 mice per group, two sections per mouse). (I) Representative images of WFA-positive cells in the mPFC (Bar = 100μm). (J) Quantification of WFA-positive cells in the mPFC (n = 3 mice per group, two sections per mouse). (K) Representative images of WFA-positive cells in mPFC subregions (Bar = 100μm). (L–M) Quantification of WFA-positive cells in (L) CG1 and (M) CG2 (n = 3 mice per group, two sections per mouse). (N) Representative image and (O) quantification of WFA-positive cells in S2 (Bar = 100 μm; n = 3 mice per group, two sections per mouse). (P) Representative image and (O) quantification of WFA-positive cells in M1 (Bar = 100 μm; n = 3 mice per group, two sections per mouse). (R) Representative image and (S) quantification of WFA-positive cells in STR (Bar = 100 μm; n = 3 mice per group, two sections per mouse). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. A total of nine mice were used.

Subregional analyses revealed significant differences in the in the hippocampal Cornu Ammonis 1 (CA1) (F(2.000, 18.59) = 25.17, P < 0.0001; Fig. 2D–E) and Cornu Ammonis 2 (CA2) (F(2, 31) = 7.357, P = 0.0024; Fig. 2F) regions, as well as in the mPFC Cingulate cortex area 1 (CG1) (F(2,30) = 11.97, P = 0.0002; Fig. 2K–L) and Cingulate cortex area 2 (CG2) (F(2, 29) = 26.88, P < 0.0001; Fig. 2M). Post hoc analysis demonstrated significant CRS-induced reductions in WFA-positive cell numbers in CA1 (t = 5.689, P < 0.0001, df = 19.18), CA2 (t = 3.490, P = 0.0044, df = 31), and CG1 (t = 3.602, P = 0.0034, df = 30), with a similar trend observed in CG2 (t = 2.396, P = 0.0698, df = 29). Ketamine treatment significantly reversed these reductions in all four regions (CA1: t = 6.100, P < 0.0001, df = 14.70; CA2: t = 3.151, P = 0.0108, df = 31; CG1: t = 4.724, P = 0.0002, df = 30; CG2: t = 7.157, P < 0.0001, df = 29).

Outside the hippocampus and mPFC, significant group differences were detected in the secondary somatosensory cortex (S2) (F(2, 27) = 5.499, P = 0.0099; Fig. 2N–O), where CRS reduced the number of WFA-positive cells relative to controls (t = 3.316, P = 0.0078, df = 27). No significant effects were observed in the primary motor cortex (M1) (F(2, 29) = 0.7268, P = 0.4920; Fig. 2P–Q) or the striatum (H(2) = 1.006, P = 0.6163; Fig. 2R–S).

Loss of PNN was independent of changes in core PNN proteins

To determine whether the observed loss of WFA-positive PNNs was associated with alterations in core PNN components, we examined aggrecan and neurocan immunoreactivity in the hippocampus and mPFC (Fig. 3A, Fig. 4A).

PNN loss is independent of changes in core PNN proteins in the hippocampus. (A) Representative images of Neurocan expression in hippocampal subregions (Bar = 50μm). (B–E) Quantification of Neurocan expression in (B) CA1, (C) CA2, (D) CA3, and (E) DG (n = 3 mice per group, 2–3 sections per mouse). (F) Representative images of Aggrecan expression in hippocampal subregions (Bar = 100μm). (G–J) Quantification of Aggrecan expression in (G) CA1, (H) CA2, (I) CA3, and (J) DG (n = 3 mice per group, 2–3 sections per mouse). (K) The main components of PNNs in the hippocampus. (L) Relative mRNA expression levels of PNN components in the hippocampus: Aggrecan (n = 6, 5, and 4), Neurocan (n = 6, 5, and 5), Brevican (n = 6, 5, and 5), Tenascin R (n = 6, 5, and 5), and Phosphacan (n = 6, 6, and 4). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. A total of 26 mice were used, and three animals were excluded from the final analysis.

CRS-induced loss is independent of changes in core PNN proteins in the medial prefrontal cortex.(A) Representative images of Neurocan expression in CG1 and CG2 (Bar = 50μm). (B–C) Quantification of Neurocan expression in (B) CG1 and (C) CG2 (n = 2 mice per group, three sections per mouse). (D) Representative images of Aggrecan expression in CG1 and CG2 (Bar = 50μm). (E–F) Quantification of Aggrecan expression in (E) CG1 and (F) CG2 (n = 2 mice per group, two sections per mouse). (G) The main components of PNNs in the mPFC. (H) Relative mRNA expression levels of PNN components in HIP: Aggrecan (n = 5, 6, and 5), Neurocan (n = 5, 4, and 6), Brevican (n = 5, 4, and 6), Tenascin R (n = 5, 5, and 6), Phosphacan (n = 4, 5, and 6). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. A total of 23 mice were used, and three animals were excluded from the final analysis.

Hippocampus

Neurocan immunoreactivity differed significantly among the three groups in all hippocampal subregions examined, including CA1 (F(2, 27) = 8.567, P = 0.0013), CA2 (F(2, 28) = 7.794, P = 0.0020), Cornu Ammonis 3 (CA3) (F(2, 30) = 11.69, P = 0.0002), and Dentate Gyrus (DG) (F(2, 29) = 8.497, P = 0.0012) (Fig. 3A–E). Compared with controls, CRS significantly reduced neurocan immunoreactivity in CA1 (t = 3.971, P = 0.0014, df = 27), CA2 (t = 3.922, P = 0.0016, df = 28), CA3 (t = 4.578, P = 0.0002, df = 30), and DG (t = 3.984, P = 0.0013, df = 29). Ketamine treatment significantly reversed these reductions in CA1 (t = 2.990, P = 0.0002, df = 27), CA3 (t = 4.143, P = 0.0008, df = 30), and DG (t = 3.080, P = 0.0135, df = 29), but not in CA2.

In contrast, aggrecan immunoreactivity was largely unchanged across hippocampal subregions. Significant group differences were observed only in CA3 (F(2, 39) = 5.302, P = 0.0092), whereas no significant effects were detected in CA1, CA2, or DG Fig. 3F–J). CRS significantly reduced aggrecan immunoreactivity in CA3 (t = 2.850, P = 0.0208, df = 39), and ketamine treatment restored its expression (t = 2.665, P = 0.0334, df = 39).

Medial prefrontal cortex

Within the mPFC, neither neurocan nor aggrecan immunoreactivity differed significantly among groups in the CG1 subregion (Fig. 4D–E). However, neurocan immunoreactivity differed significantly in CG2 (H(2) = 11.31, P = 0.0035; Fig. 4C). Dunn's post-hoc analysis showed that neurocan immunoreactivity was significantly higher in the CRS + KET group than in both the CRS + NS group (P = 0.0432) and the CON + NS group (P = 0.0033).

Similarly, Welch's ANOVA revealed significant differences in aggrecan immunoreactivity in CG2 (W(2.000, 10.89) = 9.325, P = 0.0044; Fig. 4F). Games–Howell post-hoc test showed increased aggrecan immunoreactivity in the CG2 subregion of the CRS + KET group relative to the CON + NS group (P = 0.0432).

mRNA expression of PNN components

To further investigate the molecular basis of PNN alterations, we quantified mRNA levels of major PNN components by RT-qPCR.

In the hippocampus, no significant differences were observed among groups in the expression of neurocan (F(2, 13) = 1.258, P = 0.3166), aggrecan (F(2, 12) = 1.400, P = 0.2841), brevican (F(2, 13) = 1.441, P = 0.2722), tenascin-R (F(2, 13) = 1.232, P = 0.3236), or phosphacan (F(2, 13) = 0.2754, P = 0.7636) (Fig. 3L).

Similarly, in the mPFC, expression levels of neurocan (F(2, 12) = 1.383, P = 0.2880), brevican (F(2, 12) = 1.883, P = 0.1945), tenascin-R (F(2, 13) = 2.177, P = 0.1529), and phosphacan (F(2, 12) = 0.04057, P = 0.9604) did not differ significantly among the three groups (Fig. 4G). In contrast, aggrecan mRNA expression showed a significant group effect (F(2, 13) = 6.137, P = 0.0133). Bonferroni post-hoc analysis revealed reduced aggrecan mRNA levels in the CRS + NS group compared with controls (t = 3.464, P = 0.0126, df = 13). However, ketamine treatment did not significantly restore aggrecan mRNA levels (t = 2.043, P = 0.1856, df = 13).

Effects of ketamine on the PV-positive cells in the hippocampus and mPFC of CRS mice

We next examined the effect of CRS and ketamine treatment on the number of parvalbumin (PV)-positive cells in the hippocampus and mPFC using immunohistochemistry (Fig. 5A-E).

Effect of ketamine treatment parvalbumin (PV)-positive cells in the hippocampus and medial prefrontal cortex of CRS mice. (A) Representative images of PV-positive cells in the hippocampus (Bar = 100μm). (B–E) Quantification of PV-positive cell numbers in (B) CA1, (C) CA2, (D) CA3, and (E) DG (n = 3 mice per group, three sections per mouse). (F) Representative images of PV-positive cells in the mPFC (Bar = 100μm). (G–H) Quantification of PV-positive cell numbers in (G) CG1 and (H) CG2 (n = 3 mice per group, three sections per mouse). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. A total of nine mice were used.

Hippocampus

Significant group differences were observed in all hippocampal subregions, including CA1 (H(2) = 18.99, P < 0.0001), CA2 (H(2) = 7.388, P = 0.0249), CA3 (H(2) = 11.18, P = 0.0037), and DG (H(2) = 6.998, P = 0.0302) (Fig. 5A–E). Dunn's post hoc analysis revealed that CRS significantly reduced the number of PV-positive cells compared with controls across all subregions examined (CA1, P = 0.0008; CA2, P = 0.0314; CA3, P = 0.0228; DG, P = 0.0273). Ketamine treatment significantly reversed the CRS-induced reduction in the CA3 subregion (P = 0.0062). However, no significant effects of ketamine were observed in CA1, CA2, and DG.

mPFC

No significant differences were detected in the number of PV-positive cells in the CG2 subregion (H(2) = 2.124, P = 0.3458; Fig. 5F–G). In contrast, significant differences were observed in CG1 subregion (F(2, 41) = 3.520, P = 0.0388; Fig. 5H). However, the Bonferroni post hoc analysis did not identify any significant pairwise differences between groups.

Effects of ketamine on c-Fos-positive cells in the hippocampus and mPFC of CRS mice

To assess neuronal activity, we quantified the number of c-Fos-positive cells in the hippocampus and mPFC using immunohistochemistry (Fig. 6).

Effect of ketamine treatment on c-Fos-positive cells in the hippocampus and medial prefrontal cortex of CRS mice. (A) Representative images of c-Fos-positive cells in the hippocampus (Bar = 100μm). (B–E) Quantification of c-Fos-positive cell numbers in (B) CA1, (C) CA2, (D) CA3, and (E) DG (n = 2 mice per group, two sections per mouse). (F) Representative images of c-Fos-positive cells in the mPFC (Bar = 100μm). (G–H) Quantification of c-Fos-positive cell numbers in (G) CG1 and (H) CG2 (n = 2 mice per group, two sections per mouse). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. A total of six mice were used.

Hippocampus

Significant group differences were observed in the CA3 (F(2, 15) = 4.965, P = 0.0221; Figs. 6A, 6D) and DG subregions (F(2, 17) = 22.75, P < 0.0001; Fig. 6E). In the DG, Bonferroni post-hoc analysis revealed that the CON + NS group had significantly higher numbers of c-Fos-positive cells than both the CRS + NS group (t = 4.926, P = 0.0004, df = 17) and the CRS + KET group (t = 6.640, P < 0.0001, df = 17). In the CA3, no significant difference was detected between the controls and CRS group (t = 2.638, P = 0.0559, df = 13). However, ketamine treatment significantly reduced the number of c-Fos-positive cells compared to the CRS + NS group (t = 2.706, P = 0.0488, df = 13).

mPFC

Welch's ANOVA revealed significant group differences in both CG1 (W(2.000, 10.16) = 9.490, P = 0.0047; Fig. 6F–G) and CG2 (W(2.000, 8.877) = 8.289, P = 0.0093; Fig. 6H). Games-Howell post-hoc tests demonstrated that CRS significantly increased the number of c-Fos-positive cells compared to controls in both CG1 (t = 3.903, P = 0.0173, df = 6.271) and CG2 (t = 3.460, P = 0.0304, df = 6.098). Ketamine treatment significantly reversed the CRS-induced increase in c-Fos-positive cells in CG1 (t = 2.945, P = 0.0488, df = 7.172), whereas no significant effect was observed in CG2 (t = 1.754, P = 0.2349, df = 9.766).

Effects of ketamine on the expression of PNN synthesis- and degradation-related enzymes in the hippocampus and mPFC

To explore potential mechanisms underlying PNN remodelling, we examined the expression of genes involved in PNN degradation and synthesis (Fig. 7).

Effect of ketamine treatment on the expression of PNN-synthesizing and PNN-degrading enzymes in the hippocampus and medial prefrontal cortex of CRS mice. (A) Relative mRNA expression levels of MMP2 (n = 6 mice per group), MMP9 (n = 6, 6, and 5 mice), ADAMTS1, ADAMTS4, ADAMTS5 (n = 6 mice per group) in the hippocampus. (B) Relative mRNA expression levels of CSGalNAcT1 and CSGalNAcT2 in the hippocampus (n = 6 mice per group). (C) Relative mRNA expression levels of MMP2, MMP9, ADAMTS1, ADAMTS4, ADAMTS5 in the mPFC (n = 6 mice per group). (D) Relative mRNA expression levels of CSGalNAcT1 and CSGalNAcT2 in the mPFC (n = 6 mice per group). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. A total of 18 mice were used.

Hippocampus

No significant differences were observed among the three groups in hippocampal mRNA expression levels of MMP2 (F(2, 15) = 0.7654, P = 0.4825), MMP9 (H(2) = 0.7836, P = 0.6955), ADAMTS1 (F(2, 15) = 0.3443, P = 0.7142), ADAMTS4 (F(2, 15) = 0.1615, P = 0.8523), ADAMTS5 (F(2, 15) = 1.345, P = 0.2902), CSGalNAcT1 (F(2, 15) = 0.812, P = 0.4626) or CSGalNAcT2 (F(2, 15) = 3.533, P = 0.0553) (Fig. 7A–B).

mPFC

Similarly, no significant group differences were detected in the mRNA expression of MMP9 (F(2, 14) = 3.171, P = 0.0731), ADAMTS1 (F(2, 15) = 1.838, P = 0.1933), ADAMTS4 (F(2, 15) = 0.8159, P = 0.4610), ADAMTS5 (F(2, 15) = 1.370, P = 0.2841), or CSGalNAcT1 (F(2,15) = 1.092, P = 0.3609) (Fig. 7C–D).

However, MMP2 expression differed significantly among groups (F(2,15) = 3.984, P = 0.0409). Bonferroni post-hoc analysis showed that MMP2 mRNA levels were significantly higher in the CRS + KET group than in the CON + NS group (t = 2.801, P = 0.0403, df = 15). In addition, CSGalNAcT2 expression differed significantly among groups (H(2) = 7.029, P = 0.0225). Dunn's multiple comparisons test revealed significantly lower CSGalNAcT2 mRNA levels in the CRS + KET group than in the CON + NS group (P = 0.0242) (Fig. 7D).

Effects of ketamine on the number and morphology of IBA1-positive cells in the hippocampus and mPFC of CRS mice

To evaluate the potential contribution of microglia to PNN remodelling, we quantified Iba1-positive cells and assessed microglial morphology in the hippocampus and mPFC.

Iba1-positive cell density

Significant group differences in the number of Iba1-positive cells were observed across multiple hippocampal and mPFC subregions (Fig. 8). Specifically, significant differences were observed in CA3 (F(2, 51) = 8.275, P = 0.0008), CG1 (F(2, 31) = 8.650, P = 0.0010), and CG2 (F(2, 33) = 23.32, P < 0.0001) by one-way ANOVA; in CA2 by, Welch's ANOVA (W(2, 33.17) = 8.773, P = 0.0009); and in CA1 (H(2) = 16.70, P = 0.0002), and DG (H(2) = 19.86, P < 0.0001) by Kruskal–Wallis analysis.

Effect of ketamine treatment on IBA1-positive microglia in the hippocampus and medial prefrontal cortex of CRS mice. (A) Representative images of the number of microglia per square micrometre in hippocampal subregions (Bar = 100μm). (B–E) Quantification of microglia density in (B) CA1, (C) CA2, (D) CA3, and (E) DG (n = 4 mice per group, 2–3 sections per mouse). (F) Representative images of the number of microglia per square micrometre in mPFC subregions (Bar = 100μm). (G–H) Quantification of microglial density in (G) CG1 and (H) CG2 (n = 2 mice per group, three sections per mouse). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus CON + NS. A total of 12 mice were used.

Dunn’s multiple comparisons test showed a reduced number of Iba1-positive cells in the CRS + KET group. Compared with controls, the CRS + KET group exhibited significantly fewer Iba1-positive cells in CA1 (vs. CON+NS, P = 0.0011; vs. CRS+NS, P = 0.0026) and DG (vs. CON+NS, P < 0.0001; vs. CRS+NS, P = 0.0213). A consistent reduction was also observed in the CA2 (vs. CON+NS, t = 4.104, P = 0.0007, df = 33.25; vs CRS+NS, t = 3.271, P = 0.0065, df = 36.10). Bonferroni post-hoc test further showed that Iba1-positive cells in the CRS + KET group was significantly lower than in the CON + NS group in the CA3 (t = 4.064, P = 0.0005, df = 51), CG1 (t = 2.894, P = 0.0207, df = 31), and CG2 (t = 6.815, P < 0.0001, df = 33). Moreover, the CRS + KET group also showed significant reduction of Iba1-positive cells in the CG1 (t = 4.006, P = 0.0011, df = 31) and CG2 (t = 3.783, P = 0.0019, df = 33). Additionally, the CRS + NS group displayed lower Iba1-positive cells than the controls, specifically in CG2 (t = 3.031, P = 0.0141, df = 33).

Microglial morphology

Next, we conducted microglial branching analysis. Significant group differences in the number of microglial branches in the 9–30 μm range from the soma were observed in CA1, CA2, and CA3 (all P < 0.05). In CA2, a particularly robust effect was detected at the 27-μm branch point (W(2.000, 78.76) = 42.47, P < 0.0001; Fig. 9).

Effect of ketamine treatment on microglial morphology in the hippocampus of CRS mice (A) Representative images of microglial morphology in hippocampal subregions (Bar = 50μm). (B–E) Quantification of microglia branching at various distances across (B) CA1 (n = 45 cells per group), (C) CA2 (n = 45, 45, and 40 cells), (D) CA3 (n = 45 cells per group), and (E) DG (n = 45, 37, and 43 cells). Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 CON+CRS versus CON+NS; #p < 0.05, ##p < 0.01, ###p < 0.001, ####p < 0.0001 CON+CRS versus CRS+KET; $p < 0.05, $$p < 0.01, $$$p < 0.001, $$$$p < 0.0001 CON+NS versus CRS+KET. A total of 12 mice were used.

Compared with controls, the CRS + KET group exhibited significantly fewer branches across multiple distances in CA1, CA2, and CA3 (all P < 0.05). CRS alone also reduced branching complexity at several radii, particularly in CA1 at 9 μm (P = 0.0311), 12 μm (P = 0.0024), 18 μm (P = 0.0026), 21 μm (P < 0.0001), 24 μm (P = 0.0017), 27 μm (P = 0.0087); and in CA3 at 12 μm (P = 0.0203). Furthermore, the CRS + KET group showed significantly fewer branches than the CRS + NS group at 15 μm in CA1 (P = 0.0258), at 9–30 μm in CA2 (all P < 0.05), and at 24 μm in CA3 (P = 0.0062).

In the DG, significant group differences were detected at 9 μm (H(2) = 10.06, P = 0.0065), 18 μm (H(2) = 7.448, P = 0.0241), and 24 μm (H(2) = 7.491, P = 0.0236; Fig. 9E). Dunn's post-hoc analysis showed fewer intersections in the CRS + KET group than in controls at 9 μm (P = 0.0253), 18 μm (P = 0.0193), and 24 μm (P = 0.0377). Notably, the CRS + NS group also had fewer branches than the CON + NS group at 9 μm (P = 0.0155).

Discussion

A deeper understanding of ketamine’s mechanism of action is essential for identifying pathways that contribute to its rapid and sustained antidepressant-like effects. In the present study, we investigated whether the perineuronal nets (PNNs) are involved in this process. We observed that chronic restraint stress (CRS) induced marked reductions in WFA-positive PNNs across several brain regions, with the most prominent alterations in the medial prefrontal cortex (mPFC) and hippocampus. Ketamine treatment not only alleviated depression-like behaviours but also restored PNN expression in these regions. Notably, these effects were accompanied by reduced microglial activation. Collectively, our findings support a model in which ketamine exerts sustained antidepressant-like effects not only through NMDAR systems but also through microglia-associated PNN remodelling.

Chronic stress is a major risk factor for depression and is widely used to model depression mechanisms (Luhong et al., 2024, Tafet and Nemeroff, 2016). The discovery of ketamine’s rapid and long-lasting antidepressant effects marked a major breakthrough in the treatment of psychiatric disorders (Brown et al., 2025). Ketamine is generally considered a high-affinity NMDA receptor antagonist that induces a surge of presynaptic glutamate release, leading to an AMPAR-mediated intracellular signalling cascade that increases BDNF and mTOR activation and enhances synaptic transmission and synaptogenesis (Mcintyre et al., 2021, Abdallah et al., 2018). However, accumulating evidence indicates that ketamine also possesses neuroprotective, neuroplasticity-promoting, and anti-inflammatory properties, suggesting that its therapeutic actions extend beyond direct NMDAR blockade (Zanos et al., 2018, Camargo et al., 2021). In line with previous studies (Camargo et al., 2021, Chen et al., 2024), ketamine produced robust antidepressant-like effects in CRS mice, as evidenced by increased sucrose preference and reduced immobility in the FST and TST. Importantly, locomotor activity remained unchanged across experimental groups, indicating that the behavioural improvements were not secondary to alterations in general activity levels.

PNNs act as a physical barrier that stabilises glutamatergic inputs and shields PV neurons from oxidative and metabolic damage; PNN deficiency has been reported in a variety of neurological and neuropsychiatric disorders (Arreola et al., 2021, Crapser et al., 2020a, Crapser et al., 2020b, Lisboa et al., 2024). Nevertheless, relatively little is known about how ketamine influences PNNs and PV neurons in stress-related depression models. Our findings revealed pronounced regional differences in PNN vulnerability following chronic stress. CRS significantly reduced WFA-positive cells in the hippocampus and mPFC, whereas ketamine reversed these effects. In contrast, minimal changes were observed in other regions, except the sensory cortex. Regional specificity of PNN alterations has also been reported (Matuszko et al., 2017), while the mechanism underlying this region-specificity needs further investigation.

Previous work demonstrated that PNNs mechanism may play a role in the long-term, but not acute, antidepressant effects of ketamine (Donegan J and Lodge D, 2017). However, this study was conducted in non-stressed animals and did not directly address whether ketamine could reverse depression-induced PNN abnormalities. Our findings extend these observations by demonstrating that ketamine restores PNN integrity in the brains of chronically stressed mice. Considering the key role of the medial prefrontal cortex and hippocampus in mediating the anti-depressive effect of ketamine (Carreno et al., 2016, Gomes et al., 2020), the PNN core protein abundance was further examined in these two brain regions.

Interestingly, the loss of PNN appears to occur independently of the PNN core protein. Although region-specific changes in aggrecan and neurocan immunoreactivity were observed, their expression patterns did not closely mirror the widespread reductions in WFA staining induced by CRS or the restoration produced by ketamine. This apparent dissociation suggests that a change in PNN integrity may not primarily reflect changes in core protein abundance. In support of this hypothesis, previous reports in Alzheimer's disease have shown that PNN loss is characterised by selective deglycosylation of chondroitin sulphate proteoglycans rather than degradation of their core protein (Scarlett et al., 2022, Howell et al., 2015, Lendvai et al., 2013, Morawski et al., 2010, Morawski et al., 2012). In addition, because PNNs protect PV neurons from damage (Bucher et al., 2021), we examined whether stress-induced alterations in PNNs were associated with changes in PV expression. CRS reduced the number of PV-positive neurons throughout the hippocampus, whereas ketamine restored PV expression, particularly within the CA3 and DG regions. Surprisingly, neither CRS nor ketamine substantially altered PV-positive cell number in the mPFC despite clear changes in PNN expression. These findings indicate that PNN remodelling and PV expression following CRS and ketamine treatment are not necessarily coupled and may differ across brain regions.

Because the expression of PNN components and their enzymatic degradation are activity-dependent (Dityatev et al., 2010), the effects of ketamine on PNNs may depend on the intrinsic activity in the mPFC and hippocampus. Therefore, we further examined the neuronal activation marker c-Fos (Peng et al., 2023) in these two brain regions. Interestingly, neither CRS nor ketamine significantly altered c-Fos expression in the CA1 or CA2 regions. This finding suggests that PNN remodelling is not directly associated with enhanced local neuronal activity, but rather with external inputs to these two regions. In the dentate gyrus, reduced c-Fos positive cells persisted after ketamine treatment, suggesting that the reduced local neuronal activity in DG might be associated with PNN remodelling. In contrast, CRS increased the number of c-Fos positive cells in the mPFC, and ketamine effectively reversed this elevation. These findings suggest that neuronal activity may contribute directly to PNN remodelling in the mPFC more than in the hippocampus. Taken together, our data support a region-specific relationship among neuronal activity, PN neurons, and PNN remodelling following chronic stress and ketamine treatment.

To further explore the mechanisms underlying ketamine-induced PNN remodelling, we measured the expression of genes involved in PNN synthesis and degradation in the hippocampus and mPFC. Most transcripts remained unchanged in either brain region following ketamine administration. The observed reduction in CSGalNalT2 in the mPFC was not readily consistent with increased PNN accumulation following ketamine treatment. One possible explanation is that the mRNA changes reflect compensatory responses aimed at re-establishing homeostasis after PNN recovery. Future studies should investigate the protein levels and activity of these enzymes to clarify their contribution to stress- and ketamine-induced PNN remodelling. Overall, our findings suggest that the restoration of PNNs by ketamine cannot be solely explained by changes in the expression of known PNN synthesis- or degradation-related genes.

Recent studies have shown that microglia can phagocytose PNN components under pathological conditions (Tansley et al., 2022, Liu et al., 2025). Chronic depletion of microglia increases the accumulation of PNNs (Crapser et al., 2020b) and perisynaptic ECM containing brevican in the healthy mouse brain (Strackeljan et al., 2021). Furthermore, chronic stress is known to alter microglia number, activity, and morphology, suggesting a potential link between stress-induced neuroinflammation and PNN architecture (Morphett et al., 2024). Given the well-established anti-inflammatory properties of ketamine in humans (Nikkheslat, 2021) and animals (Spencer et al., 2022, Ommati et al., 2023), its ability to reduce inflammation may contribute to its protective effects against stress. Supporting this hypothesis, ketamine treatment markedly reduced IBA1 expression in the hippocampus and mPFC subregions. Although chronic stress did not alter the number of IBA1-positive cells, we observed reduced microglia complexity in the hippocampus. These observations suggest that chronic stress may induce microglial overactivation and phagocyte PNN, both of which are blocked by ketamine. Of relevance to our concept, microglia has been suggested to be involved in the regulation of PNN (Crapser et al., 2021, Smail et al., 2023), and depletion microglia by drugs such as PLX3397(Strackeljan et al., 2021, Strackeljan et al., 2025) or PLX5622 (Zhang et al., 2025b) would lead to the recover of PNN in other animal models, such as the AD mice (Crapser et al., 2020b). It will be essential for future studies to confirm this causal relationship and whether similar mechanisms contribute to the actions of other antidepressants.

Some limitations of the present study should be acknowledged. First, the Wisteria floribunda agglutinin (WFA) marker does not bind equally to all CSPGs. Consequently, WFA staining may not fully capture the abundance and molecular diversity of PNN-enwrapped neurons (Fawcett et al., 2019). Future studies incorporating additional markers and complementary labelling approaches will provide a more comprehensive assessment of PNN alterations. Second, only male mice were included in this study. Given the well-documented sex differences in stress susceptibility, generalizability needs to be further investigated. Third, previous studies reported that repeated ketamine anaesthesia can induce PNN loss, potentially through microglia-dependent mechanisms (Matuszko et al., 2017, Venturino et al., 2021). Because PNN components and enzymatic degradation are also activity-dependent (Dityatev et al., 2010), differences in ketamine dosage, treatment duration, and the resulting neural activity patterns may account for these discrepancies. Due to the aim of the present study and the limited resource of the project, we did not further investigate the potential link between the microglia and ketamine, especially their effect on the PNNs, which needs additional exploration. Finally, pharmacological interventions to directly manipulate PNN were not performed. The lack of selective agonists and antagonist agents limits the ability to establish causal relationships. These scientific questions warrant further investigation.

In summary, the chronic restraint test induced region-specific alterations in PNNs, mainly in the hippocampus and mPFC. Ketamine treatment reversed these stress-induced abnormalities, potentially via microglia-mediated PNN remodelling, without altering major PNN synthesis- and degradation-related genes. Advancing our understanding of how ketamine regulates PNN and the antidepressant effects may facilitate the identification of novel therapeutic targets with sustained antidepressant actions.

Authorship statement

Jingwen Yang, Ting Huang, and Meng Wang performed the behavioural tests, biochemical analyses, and histological experiments. Yingrui Xu conducted the behavioural experiments. Xuan Li, Wenyu Cao, and Changqi Li provided resources. Huamao Zhou and Liyan He provided resources, directed the experimental design, and revised the manuscript.

CRediT authorship contribution statement

Huamao Zhou: Conceptualization, Funding acquisition, Investigation, Writing – original draft, Writing – review & editing. Liyan He: Funding acquisition, Resources, Writing – review & editing. Changqi Li: Resources. Wenyu Cao: Funding acquisition, Resources. Yingrui Xu: Methodology. Xuan Li: Resources. Meng Wang: Methodology, Project administration. Jingwen Yang: Methodology, Project administration, Writing – original draft. Ting Huang: Project administration, Resources, Writing – original draft.

Declaration of Competing Interest

The authors declare that they have no conflicts of interest or any known competing financial interests or personal relationships that may bias the performance or interpretation of the work reported in this paper.