Cannabinoid use generalizes stress responses: involvement of astrocyte plasticity and activation of matrix metalloproteinases in the nucleus accumbens core

Abstract

The rising legal acceptance of cannabis and the high comorbidity between cannabis use disorder (CUD) and post-traumatic stress disorder (PTSD) highlight the importance of understanding how stress and cannabis influence the brain. We recently discovered that cannabinoid use promotes two PTSD-like symptoms: avoidance coping behaviors and the generalization of stress-coping responses to a neutral stimulus. Here, we used in vivo zymography and confocal microscopy to examine how stress and cannabinoid use influence multipartite synaptic plasticity. Specifically, we assessed astroglial plasticity, Synapsin-I density, and matrix metalloproteinases (MMP-2,9) activity, in the nucleus accumbens core (NAcore). For this purpose, rats were restrained for 2 h and simultaneously exposed to an odor; the stress-conditioned stimulus (stress-CS). Three weeks later, rats were exposed to cannabinoid vapor (delta9-tetrahydrocannabinol+cannabidiol; THC + CBD) for 5 days, self-administered THC + CBD (i.v.) for 10 days, followed by 10 days of abstinence. We then evaluated the effect of stress-CS or neutral odor (NS) on coping strategies in a defensive burying task. We found that THC + CBD generalized stress responses to the NS, associated with astrocyte retraction from synapses and a decrease in Synapsin-I density. THC + CBD pretreatment promoted avoidant coping during stress-CS exposure, activated MMP-2,9, re-associated astrocytes with synapses, increased Synapsin-I density, and caused astrocyte atrophy. By inhibiting MMP-2,9, we found that stress-CS-induced plasticity required MMP-2,9 activation. MMP-2 inhibition also restored active coping behaviors during stress-CS exposure. Surprisingly, these neuroadaptations only occurred in males. Overall, these findings suggest a potential role for MMPs and astrocytes in the changes produced by THC + CBD use in responding to a stress-CS.

Introduction

For certain individuals, a single traumatic event can lead to trauma and stress-related disorders, including post-traumatic stress disorder (PTSD) [1]. Cannabis use disorder (CUD) is often comorbid with PTSD and is the most prevalent co-occurring condition among veterans diagnosed with PTSD [2]. Recent longitudinal studies demonstrate that cannabis use is associated with higher PTSD symptom severity, including greater traumatic thought intrusions [3, 4] and depressive symptoms [5]. Consistent with this, patients with comorbid PTSD and CUD report higher PTSD symptom severity and poorer treatment outcomes relative to those without CUD [6, 7]. Despite this evidence, PTSD is often listed as a qualified medical condition for the use of cannabis in the United States [8, 9]. As of 2025, 31 states listed cannabis as a treatment for PTSD [10]. Consequently, the increasing use of cannabis for self-medication of PTSD and the comorbid PTSD and CUD underscores the need to identify the neurobiological mechanisms through which cannabis use alters and potentially exacerbates the symptoms associated with PTSD.

Using a stress cue-conditioned (stress-CS) animal model of PTSD, we recently demonstrated that an acute stress potentiates cannabinoid consumption in male rats [11]. Additionally, we discovered that cannabinoid use promotes two significant PTSD-like symptoms: generalization of stress-coping responses to a neutral stimulus not previously associated with stress, and engaging in avoidance coping behaviors when faced with a conditioned stress odor [11]. Importantly, these findings differ from those observed after cocaine and heroin self-administration, where rats exposed to a stress-CS engage in active coping behaviors, without generalizing to an odor not associated with acute stress experience [12–14]. Finally, the cannabinoid-induced behavioral changes were accompanied by reduced spine density in the nucleus accumbens core (NAcore), a brain region associated with stress and addiction, and a further decrease in spine head diameter after exposure to the stress-CS [11]. This suggests that cannabinoids may be altering how environmental stimuli elicit synaptic plasticity.

Nonneuronal components of the neuropil adjacent to synapses, including astroglia and extracellular matrix (ECM), form what is termed the multipartite synapse and function together with pre- and postsynapses to regulate synaptic plasticity [15, 16]. For instance, astrocytes regulate synaptic plasticity by surrounding synapses, thereby providing them with metabolic support, gliotransmitters, and maintaining ion homeostasis [16]. In parallel, destabilizing the ECM via proteolytic activity produced by matrix metalloproteinases (MMPs) is associated with neuronal and astroglial plasticity. Both environmental stressors and addictive drugs alter MMP activity, astroglial morphology and signaling, and contribute to the development of stress- and drug-related disorders [17]. However, the effects of cannabis use, either independently or in conjunction with stress, on the nonneuronal components of multipartite synapses remain unknown. The objective of this study was to assess whether the interaction between stress and cannabinoid use influences components of multipartite synapses, including astroglial plasticity, synapse density, and MMP activity in the NAcore. For this purpose, following an acute restraint stress, astrocytes were labeled in rats that were trained to self-administer delta9-tetrahydrocannabinol+cannabidiol (THC + CBD) [18]. After assessing how cannabinoid use affected coping strategies through a defensive burying task (DBT), we employed confocal microscopy, in vivo zymography, and digital rendering techniques to quantify astrocyte morphology, the colocalization of astrocytes with the presynaptic marker Synapsin-I, and the gelatinolytic activity of MMP-2 and MMP-9 in the NAcore.

Methods

Animals

Male and female Long-Evans rats (~250 g; Charles River Laboratories) were double-housed according to sex on a 12-h light/dark cycle with food and water available ad libitum. Experiments occurred in the light cycle.

Acute stress and odor pairing

Rats were placed in a restraining device for 2 h and paired with an odor (lemon or sandalwood; Sun essential, Phoenix, AZ, USA) that became the stress-CS. Control rats were exposed to the same odor in the home cage for 2 h (Sham-CS). The odor that was not stress- or sham-paired was used during the DBT as a neutral odor control (stress-NS). Two weeks after the stress procedure, rats received indwelling catheters as previously described [18].

Viral labeling

After catheter implantation, rats were implanted with bilateral guide cannulae into the NAcore (AP: 1.5 mm, ML: ± 1.7 mm, DV: −5.5 mm). AAV5/GFAP-hM3dq-mCherry (University of Zurich) was injected into the left hemisphere (1.0 μl, 0.15 μl/min, 5 min diffusion) to label astrocytes. The microinjectors are 2 mm longer than the guide cannulae to reach the target.

THC+CBD self-administration

Following surgery, rats underwent 5 days of vapor exposure to a THC + CBD mixture in a 10:1 ratio, as previously described [18]. Rats were then trained to self-administer THC + CBD on a fixed ratio 1 schedule, during 90 min sessions for 10 consecutive days. Active lever presses delivered 0.4 + 0.04 mg/kg THC + CBD and were paired with cues (light+tone) followed by a 20-s timeout. During the last 5 days, once self-administration is well-established, a discrimination index was computed as active lever presses−inactive lever presses/active lever presses+inactive lever presses, with values ranging from 0 (no discrimination) to 1 (perfect discrimination). Another cohort of rats was trained to self-administer the vehicle (1% ethanol). Following self-administration, animals underwent 10 days of abstinence to ensure the absence of THC metabolites and avoid spontaneous cannabinoid withdrawal symptoms during the DBT [19].

Defensive burying task

DBT was performed as previously described [11]. Briefly, bedding was placed in one half of the standard home cage, opposite the odor. Rats were placed facing away from the odor on the bedding side of the cage, and behavior was recorded for 15 min using Ethovision XT software (Leesburg, VA, USA). Active (burying) and avoidant (escape, immobility and grooming) coping mechanisms were quantified.

In vivo zymography

In vivo zymography was used to quantify MMP-2,9 activity as previously described in detail [20]. FITC-gelatin was microinjected into the NAcore right hemisphere (1.5 µl, 0.5 µl/min, 3 min diffusion) immediately before DBT. Rats were then perfused with 4% PFA immediately after the DBT and brains were sliced at 100 µm. Only slices containing the injection track and the anterior commissure on the same frame were analyzed. The fluorescence intensity linearly reflects the gelatinolytic activity of MMP-2,9 [21]. MMP-2 (444288, MilliporeSigma) and MMP-9 (444278, MilliporeSigma) inhibitors were dissolved in 1% DMSO and bilaterally microinjected before FITC-gelatin in a separate rat cohort as previously described [22].

Immunohistochemistry

Left hemisphere slices, containing the GFAP-mCherry virus, were permeabilized for 15 min in PBS with 0.2% Triton X-100 (PBST), then blocked with 2% PBST containing 5% normal goat serum. Rabbit anti-synapsin-I antibody (1:1000, ab64581, Abcam) was incubated for 72 h at 4 °C. Slices were washed with 0.2% PBST, then incubated with Alexa Fluor 405(1:1000, ab175652, Abcam) for 72 h at 4 °C.

Confocal microscopy

Astrocyte images were acquired using a Leica SP5 with a 63X oil-immersion objective lens at 16-bit resolution with a 1-μm step size. Images were then imported into Imaris software (version 9.7.0). Astrocytes were selected only if the entire cell was imaged within the section and its borders remained distinct from neighboring labeled cells. The mCherry signal was used for 3-dimensional reconstruction of astrocytes. Each rendered astrocyte was used as a region of interest (ROI) for colocalization analysis with Synapsin-I. The colocalization channel displays the results as the percentage of the ROI overlapping with Synapsin-I puncta. To determine the Synapsin-I density, we normalized the thresholded signal intensity to the volume of the frame from which it was acquired. Imaging and analysis were performed in a blinded manner.

Statistics

Data were analyzed with GraphPad Prism 10 and Python with the statmodels library. Self-administration data were analyzed using two-way repeated measures ANOVAs. Behavioral and in vivo zymography data were analyzed using two-way ANOVAs with Tukey’s post hoc tests. For astroglial and Synapsin-I measures, a D’Agostino & Pearson test was initially used to assess the normality of each group. A Linear Mixed Model with a Wald chi-square post hoc comparison was used for non-normally distributed treatment groups (see Table S1). Coping strategies, astroglial and Synapsin-I distributions were analyzed by using separate Chi-square tests followed by a Bonferroni correction for multiple comparisons. Refer to the supplementary methods and Tables S1 and S2 for detailed statistical and methodological information.

Results

Acute stress potentiated THC+CBD use in male rats

Rats were restrained for 2 h and simultaneously exposed to an odor. 3 weeks after the stress, rats were trained to self-administer THC + CBD or its vehicle for 10 days, followed by 10 days of abstinence (Fig. 1A). In THC + CBD-trained rats, stressed males pressed more on the active lever (Fig. 1B) and consumed more THC + CBD than sham rats (Fig. 1C). Both stressed and sham rats showed equivalent discrimination between the active and inactive levers (Fig. 1D). No differences between stress and sham groups were observed in vehicle-trained rats (Fig. 1E–G).

Fig. 1. Acute restraint stress potentiated THC+CBD use in males.

A Experimental timeline outlining THC + CBD self-administration, abstinence and the defensive burying task, used in Figs. 1, 2–5. P represents the postnatal weeks. A total of 6 treatment groups were used to discern the effects of stress vs sham pretreatment, THC + CBD vs vehicle self-administration, and stress-conditioned vs neutral odors. This design allowed us to interpret associations among stress, CS, and THC + CBD on MMP activity, astroglial morphology, and the behavioral components of defensive burying. B Acute restraint stress heightened active lever pressing during THC + CBD self-administration (two-way repeated measure ANOVA stress: F_1,21 = 4.54, p = 0.045; days: F_9,189, p = 0.002, ƞ² = 0.08). C Stressed rats consumed more THC + CBD compared to sham animals (two-way repeated measure ANOVA stress: F_1,21 = 5.404, p = 0.03, ƞ² = 0.1). D The sham and stress groups showed no differences in their ability to distinguish between the active (THC + CBD infusion) and inactive (no consequence) levers during the final 5 days of self-administration, which corresponds to a stage where self-administration is well established. The dotted line represents a 2:1 ratio of active to inactive. E Acute stress did not affect the active lever presses during vehicle self-administration. F Sham and stress groups consumed the same amount of vehicle. G No difference was observed between sham and stress rats on the discrimination index during vehicle self-administration. See table S2 for complete analysis. Data are shown as mean ± SEM. Number of animals is expressed in brackets. *p < 0.05, comparing stress to sham.

In female THC + CBD-trained rats, no differences were measured between groups on the active lever presses and THC + CBD intake (Figure S1A-B). However, stressed rats had a higher discrimination index than sham rats (Figure S1C), indicating that the acute stress may increase the reinforcing value of the THC + CBD in females. No differences between stress and sham groups were observed in vehicle-trained rats (Figure S1D-F).

THC+CBD promoted avoidant coping strategies in male rats

Following the abstinence period, we used DBT to evaluate the effect of THC + CBD on active (burying) and avoidant (escape, immobility and grooming) coping strategies. To this end, stressed rats were exposed to either the stress-CS odor previously associated with the stressful event or a neutral odor stimulus (NS) for 15 min. Non-stressed rats were exposed to the odor previously associated with the sham experience. Regarding active coping, in vehicle-trained male rats, the stress-CS caused an increase in burying compared to the sham group. Conversely, in THC + CBD-trained rats, the stress-NS group exhibited greater burying than the sham group. Furthermore, compared to vehicle rats, THC + CBD self-administration elevated burying in the stress-NS group while decreasing it in the stress-CS group (Fig. 2A). Regarding avoidant coping strategies, we found that THC + CBD self-administration induced a strong escape behavior across stress conditions compared to vehicle rats (Fig. 2B). Conversely, the percent time spent immobile was reduced after THC + CBD (Fig. 2C). Moreover, stress-CS significantly increased grooming in vehicle-trained rats, while both stressed groups (CS and NS) exhibited a higher grooming frequency than the sham group in THC + CBD-trained rats (Fig. 2D). When comparing the distribution of coping strategies between groups (Fig. 2E), we found no difference among the stress-NS and sham-CS groups in vehicle-trained rats, while the stress-CS promoted an active coping strategy by increasing burying behavior. THC + CBD abstinence promoted avoidant coping in the sham group, while a neutral odor stimulus facilitated both active and avoidant coping in stressed rats. Also, exposure to the stress-CS after THC + CBD further changed stress behaviors compared to the sham and stress-CS veh groups. Combined, these data indicate that THC + CBD use caused male rats exposed to an earlier stressor to generalize stress responses to neutral stimuli and worsen stress responses to conditioned stimuli.

Fig. 2. THC+CBD generalized stress responses in males.

A Stress-CS increased the time spent burying in vehicle-trained rats, while the Stress-NS led to an increase in the time spent burying in THC + CBD-trained rats (interaction: F_2,39 = 14.77, p < 0.0001, ƞ² = 0.39). B THC + CBD use generalized the escape behavior across all stress conditions (drug: F_1,39 = 41.35, p < 0.0001, ƞ² = 0.49). C THC + CBD reduced immobility in all three conditions (drug: F_1,39 = 4.842, p = 0.0338, ƞ² = 0.1). D Stress-CS enhanced grooming frequency in vehicle-trained rats, whereas both Stress-NS and Stress-CS increased grooming in THC + CBD-trained rats (stress: F_2,39 = 9.835, p = 0.0003, ƞ² = 0.3). Number of rats is expressed in C. Data are shown as mean ± SEM. Each dot in bar represents one rat. * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group, using a Tukey post hoc. E Donut charts showing the distribution of the different coping strategies within each treatment. Stress-CS changed coping strategies in vehicle- (X²₍₃₎ = 34.06, p < 0.0001 versus Sham-CS veh) and THC + CBD-trained rats (X²₍₃₎ = 34.06, p < 0.0001 versus Sham-CS THC + CBD). However, THC + CBD promoted avoidant coping strategies during stress-CS exposure (X²₍₃₎ = 106.5, p < 0.0001 versus Stress-CS veh). THC + CBD altered coping distribution in non-stressed rats (X²₍₃₎ = 974.4, p < 0.0001 versus Sham-CS veh). THC + CBD modified stress responses during exposure to the neutral stimulus (X²₍₃₎ = 70.45, p < 0.0001 versus stress-NS veh; X²₍₃₎ = 113.2, p < 0.0001 versus Sham-CS THC + CBD). * p < 0.05, compared to Sham-CS within each drug group; # p < 0.05, comparing THC + CBD to vehicle within each stress/sham group using a Chi-square.

Surprisingly, in females, neither the stress nor the THC + CBD significantly affect coping strategies in the DBT (Figure S2).

Abstinence from cannabinoids reduced astrocyte-Synapsin-I co-registration but promoted their re-association after stress-CS exposure in male rats

To determine how stress and cannabinoids influence the plasticity of NAcore astrocytes, astroglia were labeled with a membrane-bound fluorescent reporter. Rats were perfused immediately after DBT, and NAcore slices labeled with the presynaptic marker Synapsin-I. To quantify astrocyte-Synapsin-I co-registration, each astrocyte was digitally rendered, and the co-labeling of the astrocyte surface with Synapsin-I was quantified (Fig. 3A). Consistent with our previous report [23], we found that acute stress (stress-NS) induced a constitutive reduction in astrocyte-Synapsin-I association, while exposure to the stress-CS resulted in a re-association between astrocytes and Synapsin-I in vehicle-trained male rats. Abstinence from THC + CBD decreased astrocyte-Synapsin-I association in the sham group compared to vehicle rats. In stressed THC + CBD-trained rats, exposure to the NS further decreased the association between astrocytes and Synapsin-I relative to sham, while re-exposure to the stress-CS enhanced astrocyte-Synapsin-I co-registration (Fig. 3B). To examine how earlier stress and cannabinoid use influence astrocyte-Synapsin-I distribution, we employed cumulative frequency distribution to create three equal groups (low, mid, and high colocalization) within the sham-CS vehicle group. The multiple Chi² analysis revealed that exposure to the stress-CS did not fully restore the astrocyte-Synapsin-I association in vehicle-trained rats (Fig. 3C). Indeed, the astrocytes having less co-registration with Synapsin-I constituted a larger population in the stress-CS group than in the sham-CS group. Additionally, after stress-CS re-exposure, the THC + CBD group exhibited a greater proportion of astrocytes with high Synapsin-I colocalization compared to vehicle rats.

Fig. 3. THC+CBD use reduced the astroglial co-registration with synapses in NAcore but potentiated the astrocyte-synaptic association after the stress-CS exposure in males.

A Representative 63X confocal images of NAcore astrocytes for stress condition in Vehicle- (left panel) and THC + CBD-trained rats (right panel). (1) Merged image of a GFAP-hM3dq-mCherry astrocyte (red) with the presynaptic marker Synapsin-I (blue). (2) Immunolabeled Synapsin-I channel used for the quantification of Synapsin-I density. (3) Digital rendering of the astrocyte in panel 1. Scale bar = 10 µm. (4) Digitally isolated Synapsin-I puncta that colocalized with the astroglial volume (white). B Abstinence from THC + CBD constitutively reduced astrocyte-Synapsin-I association, while re-exposure to the stress-CS strongly increased the colocalization between astrocytes and Synapsin-I (X²₍₅₎ = 50.37, p < 0.0001, R²_m = 0.3351). Data are shown as mean ± SEM. N shown in legend as astrocytes/animals. * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group. C Stress-CS increased the proportion of astrocytes with high Synapsin-I colocalization in THC + CBD-trained rat (X²₍₂₎ = 19.20, p < 0.0001 versus Stress-CS veh). * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group using a Bonferroni correction. See table S2 for complete analysis. D The negative correlation between the astrocyte-Synapsin-I colocalization and grooming frequency in vehicle-trained rats (Spearman r = −0.476, *p = 0.0435) disappeared following THC + CBD use. E Negative correlation between the astrocyte-Synapsin-I colocalization and time spent burying in THC + CBD-trained rats (Pearson r = −0.444, *p = 0.0369), not in vehicle-trained rats. F Acute stress and THC + CBD abstinence (Stress-NS) reduced the Synapsin-I density (X²₍₅₎ = 25.9573, p = 0.0001, R²_m = 0.2369). Data are shown as mean ± SEM. N shown in legend as astrocytes/animals. * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group. G The stress-CS THC + CBD demonstrated a significantly higher proportion of high Synapsin-I density compared to Sham-CS THC + CBD (X²₍₂₎ = 17.5, p = 0.0002) and to Stress-CS vehicle (X²₍₂₎ = 21.82, p < 0.0001). * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group using a Bonferroni correction.

Astrocyte-Synapsin-I association negatively correlated with grooming in vehicle-trained males (Fig. 3E, left). This correlation disappeared after THC + CBD use (Fig. 3D, right). Furthermore, while no correlation existed between astrocyte-Synapsin-I association and burying in vehicle-trained rats (Fig. 3E, left), a negative correlation appeared after THC + CBD use (Fig. 3E, right). This suggests that using cannabinoids alters the astroglial function in managing coping strategies.

In females, exposure to the stress-CS increased the astrocyte-Synapsin-I co-registration in both vehicle- and THC + CBD-trained rats (Figure S3A). Compared to vehicle rats, astrocytes near Synapsin-I were more abundant in the stress-NS group, while stress-CS exposure nearly eliminated astrocytes at mid-distance in THC + CBD females (Figure S3B). Moreover, in THC + CBD-trained females, astrocyte-Synapsin-I association correlated positively with burying behavior and negatively with grooming (Figure S3C-D).

Combination of acute stress and cannabinoid use reduced Synapsin-I density in NAcore

Chronic stress is known to reduce levels of synaptic markers, such as Synapsin-I [24]. We previously found that neither acute stress nor stress-CS exposure altered Synapsin-I density in the NAcore of male rats [23]. Nevertheless, the impact of acute stress followed by cannabinoid use on Synapsin-I remains unknown. Similar to our prior report, we found that Synapsin-I density remained unchanged in vehicle-trained males, across stress conditions (Fig. 3F). Likewise, no difference was found between sham-CS groups. These findings suggest that the changes in astrocyte-Synapsin-I colocalization observed in vehicle-trained rats and during abstinence from THC + CBD result solely from astrocytic motility. However, the combination of prior stressful experience and THC + CBD abstinence lowered Synapsin-I density in stress-NS rats, while exposure to the stress-CS restored Synapsin-I density. The multiple Chi² analysis revealed that THC + CBD stress-CS group showed a higher proportion of high Synapsin-I density than the THC + CBD sham-CS and vehicle stress-CS groups (Fig. 3G). This indicates that the potentiation of astrocyte-synaptic contacts induced by the drug following stress-CS exposure observed in Fig. 3B could partly result from increased neuronal synapses associated with astroglia.

Unlike males, the interaction between prior acute stress and cannabinoid use increased Synapsin-I density in female NAcore (Figure S3E). Stressed vehicle females, exhibited a higher proportion of low Synapsin-I density than the sham-CS group. These effects are reversed after THC + CBD use (Figure S3F).

Stress-CS induced astrocyte atrophy after cannabinoid use in male rats

Chronic stress causes astrocyte shrinkage in various brain areas [25]. Previously, we found that neither stress nor stress-CS influence astrocyte structure in the NAcore of males [23]. We replicated these data in vehicle males and demonstrated that after THC + CBD use, the stress-CS reduced both the surface area and the volume of astrocytes (Fig. 4A, C). Moreover, in Sham rats, THC + CBD use increased the subpopulation of astrocytes with smaller surface areas compared to Sham vehicle rats (Fig. 4B). Additionally, in vehicle-trained rats, the stress-CS group exhibited a higher proportion of astrocytes with larger volumes than the sham group (Fig. 4D).

Fig. 4. THC+CBD use induced astrocyte atrophy after the stress-CS exposure in males.

A Stress-CS decreased the surface area in THC + CBD-trained rats (X²₍₅₎ = 26.15, p = 0.0001, R²_m = 0.184). Data are shown as mean ± SEM. N shown in legend as astrocytes/animal. * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group. B THC + CBD use increased the subpopulation of astrocytes with small surface area in Sham rats (X²₍₂₎ = 14.33, p = 0.0008 versus Sham-CS Veh), which was further augmented by stress-CS exposure (X²₍₂₎ = 15.19, p = 0.0005 versus Sham-CS THC + CBD). * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group using a Bonferroni correction. C Stress-CS decreased the volume in THC + CBD-trained rats (X²₍₅₎ = 32.86, p < 0.0001, R²_m = 0.2228). Data are shown as mean ± SEM. N shown in legend as astrocytes/animals. * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group. D Stress-CS increased the subpopulation of astrocytes with large volume in vehicle-trained rats (X²₍₂₎ = 13.33, p = 0.001 versus Sham-CS Veh), and the effect of stress-CS was reversed following THC + CBD use (X²₍₂₎ = 64.4, p < 0.0001 versus Stress-CS Veh). * p < 0.05 compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group using a Bonferroni correction. E THC + CBD changed a positive correlation between astrocyte-Synapsin-I colocalization and surface area (left; Pearson r = 0.486, p = 0.039) into a negative one (right; Pearson r = −0.512, *p = 0.0177). F Astrocyte-Synapsin-I colocalization negatively correlated with volume following THC + CBD use (Pearson r = −0.612, *p = 0.0039).

Cannabinoid use in males transformed a positive correlation between surface area and astrocyte-Synapsin-I colocalization in vehicle-trained rats into a negative correlation (Fig. 4E). Additionally, in THC + CBD-trained rats, the volume showed a negative correlation with astrocyte-Synapsin-I colocalization, unlike in vehicle rats (Fig. 4F). In summary, these data suggest that exposure to a stress-CS interacts with cannabinoid abstinence to induce an increase in both astroglial association with synapses and astrocyte atrophy.

In females, the acute stress induced a constitutive decrease in astrocyte surface area and volume in both vehicle- and THC + CBD-trained rats. Re-exposure to stress-CS restored astrocyte morphology in vehicle females, but THC + CBD prevented this effect (Figure S4A-D).

MMP-2,9 activity is increased during cannabinoid abstinence and is potentiated following the stress-CS exposure in male rats

MMP- 2 and −9 are key components of the ECM and multipartite plasticity [16, 26]. Both addictive drugs and stress exposure alter the activity of MMP-2 and MMP-9 [17]. Using in vivo zymography we quantified the gelatinolytic activity of MMP-2,9 during the DBT (Fig. 5A) [20]. Abstinence from THC + CBD increased NAcore MMP-2,9 activity regardless of the stress condition in males (Fig. 5B). Moreover, stress-CS exposure increased the MMP-2,9 activity in both vehicle and THC + CBD males. However, THC + CBD use potentiated the MMP-2,9 activity compared to vehicle rats.

Fig. 5. THC+CBD use potentiated the MMP-2,9 activity induced by the stress-CS in males.

A Representative micrograph of in vivo zymography assay from each treatment group. The dashed line delineates the anterior commissure (ac) and the injection track that were masked for quantification. Scale bar = 200 µm. B THC + CBD abstinence increased MMP-2,9 activity across all stress conditions (drug: F_1,31 = 54.84, p < 0.0001, ƞ² = 0.33). Stress-CS increased the MMP-2,9 activity in NAcore of both vehicle- and THC + CBD-trained rats (stress: F_2,31 = 33.03, p < 0.0001, ƞ² = 0.39), but THC + CBD use exacerbated the MMP-2,9 activity (interaction: F_2,31 = 7.686, p = 0.0019, ƞ² = 0.09). Data are shown as mean ± SEM. * p < 0.05, compared to Sham-CS within each drug group; # p < 0.05 comparing THC + CBD to vehicle within each stress/sham group, using a Tukey post hoc. C THC + CBD shifted a positive correlation between burying and MMP-2,9 activity (left; Pearson r = 0.485, p = 0.03) into a negative correlation (right; Pearson r = −0.456, *p = 0.03). D MMP-2,9 activity positively correlated with escape behavior in THC + CBD-trained rats (Spearman r = 0.54, *p = 0.027). E MMP-2,9 activity positively correlated with astrocyte-Synapsin-I colocalization (Pearson r = 0.75, *p = 0.0049), and F) negatively correlated with astroglia surface area following THC + CBD use (Pearson r = −0.547, *p = 0.033).

Cannabinoid use reversed a positive correlation between MMP-2,9 activity and burying behavior in vehicle-trained rats into a negative correlation (Fig. 5C). Moreover, MMP-2,9 activity was positively correlated with the escape behavior in THC + CBD-trained, not vehicle rats (Fig. 5D). These data indicate that cannabinoid use modifies how MMP-2,9 regulates coping strategies, specifically by inducing a shift from active to avoidant coping. Additionally, after cannabinoid use, MMP-2,9 activity positively correlated with astrocyte-Synapsin I colocalization, while negatively correlating with astrocyte surface area (Fig. 5E, F), suggesting that MMP-2,9 regulates astroglial plasticity.

Neither stress nor the THC + CBD alter the MMP-2,9 activity in females (Figure S5A). However, similar to males, MMP-2,9 activity showed a positive correlation with astrocyte-Synapsin-I colocalization and a negative correlation with astrocyte volume after cannabinoid use (Figure S5B-C).

Both MMP-2 and MMP-9 mediated the astrocyte-Synapsin-I association, while MMP-2 influenced astrocyte morphology

To determine whether the changes in astroglial plasticity induced by the stress-CS in males are preceded by changes in the MMP-2,9 catalytic activity, we examined the effect of MMP-2 and MMP-9 inhibitors on stress-CS-induced astroglial plasticity. To this end, during THC + CBD abstinence, a separate cohort of stressed males received bilateral microinjections of inhibitors (0.1 nmol) or their vehicle (1% DMSO) in NAcore, followed by a unilateral microinjection of the FITC-gelatin before stress-CS exposure (Fig. 6A).

Fig. 6. MMP-2 modulated the multipartite synaptic plasticity following THC+CBD use.

A Experimental timeline outlining THC + CBD self-administration, abstinence and the defensive burying task. B MMP-2i and MMP-9i did not alter the MMP-2,9 activity. C Left panel: Both MMP-2i and MMP-9i reduced the astrocyte-Synapsin-I colocalization (X²₍₂₎ = 18.08, p < 0.0001, R²_m = 0.2039). Data are shown as mean ± SEM. N shown in legend as astrocytes/animals. Right panel: MMP-9i increased the subpopulation of astrocytes located farther from Synapsin-I (X²₍₂₎ = 18.13, p = 0.0001 versus DMSO), while MMP-2i nearly eliminated astrocytes located near Synapsin-I (X²₍₂₎ = 10.51, p = 0.0052 versus MMP9i). * p < 0.05 compared to DMSO; # p < 0.05 comparing 9i to 2i. D Left panel: MMP-2i reduced the Synapsin-I density (X²₍₂₎ = 7.58, p = 0.022, R²_m = 0.08). Data are shown as mean ± SEM. N shown in legend as astrocytes/animals. Right panel: MMP-2i increased the proportion of astrocytes surrounded by low Synapsin-I density compared to DMSO (X²₍₂₎ = 24.72, p < 0.0001) and MMP-9i (X²₍₂₎ = 17.88, p = 0.0001). * p < 0.05 compared to DMSO; # p < 0.05 comparing 9i to 2i. E Left panel: MMP-2i increased astroglia surface area (Wald X²₍₁₎ = 4.79, p = 0.028, versus MMP-9i). Right panel: MMP-2i significantly reduced the proportion of astrocytes with small surface area compared to DMSO (X²₍₂₎ = 24.75, p < 0.0001). * p < 0.05 compared to DMSO; # p < 0.05 comparing 9i to 2i. F Left panel: MMP-2i increased astroglia volume (Wald X²₍₁₎ = 5.32, p = 0.021, versus MMP-9i). Data are shown as mean ± SEM. N shown in legend as astrocytes/animals. Right panel: MMP-2i nearly eliminated the proportion of small-volume astrocytes compared to DMSO (X²₍₂₎ = 23.3, p < 0.0001). * p < 0.05 compared to DMSO; # p < 0.05 comparing 9i to 2i. G Both MMP-2i (X²₍₂₎ = 10.6, p = 0.005 versus DMSO) and MMP-9i (X²₍₂₎ = 10.32, p = 0.0058 versus DMSO) changed coping responses during exposure to the stress-CS. Moreover, MMP-2i also changed the coping distribution compared to MMP-9i (X²₍₂₎ = 96.79, p < 0.0001).

Although only nonsignificant reductions in gelatinolytic activity were produced by either MMP inhibitor (Fig. 6B), both MMP-2 and −9 inhibitors reduced the astrocyte-Synapsin-I colocalization (Fig. 6C left and S6A). Furthermore, the MMP-2 inhibitor nearly eliminated astrocytes found near Synapsin-I (Fig. 6C right). A likely contributor to the reduction in astrocyte-synaptic contact was markedly reduced Synapsin-I density after treatment with the MMP-2, but not MMP-9 inhibitor (Fig. 6D). Finally, the MMP-2 inhibitor reversed astrocyte atrophy by elevating both surface area and volume compared to MMP-9i, while notably decreasing the small astrocyte subpopulation relative to DMSO (Fig. 6E, F).

While MMP inhibition provides direct causal evidence for contributing to the stress-CS induced adaptations in astrocytes of THC + CBD male rats, the MMP inhibitors also altered the behavioral profile of the stress-CS rats (Fig. 6G and S6B-E), indicating that MMP-2,9 also contributes to the behavioral adaptations produced by THC + CBD in stress-CS coping behaviors. Compared with DMSO microinjection, the MMP-9 inhibitor notably reduced burying behavior (Fig. 6G and S6B). The coping strategies also significantly differed between the MMP-2 and MMP-9-treated groups, with the MMP-2 inhibitor significantly reducing grooming (Fig. 6G and S6E).

Discussion

We recently demonstrated that cannabinoid use, after a single stressful event, promotes two PTSD-like symptoms: generalization of stress responses to neutral stimuli and avoidant coping strategies following conditioned stress exposure [11]. To explore the neuroadaptations behind these changes, we used in vivo zymography and confocal microscopy to assess how stress and cannabinoid use affect multipartite synaptic plasticity in the NAcore, a brain region linked to stress and addiction cues responding [27]. We made four primary discoveries: 1) cannabinoid abstinence, without any stressful event, reduced astroglial contact with synapses and increased MMP-2,9 activity, 2) a prior stressful event reduced Synapsin-I density during cannabinoid abstinence, 3) exposure to a stress-CS after cannabinoid use led to avoidant coping strategies by activating MMP-2,9 in the NAcore, resulting in the re-association of astrocytes with synapses and astrocyte atrophy, and 4) these multipartite adaptations by stress and cannabinoids occurred only in males.

THC+CBD disrupts coping strategies in males

In line with our previous report, we found that THC + CBD use generalized stress responses to a neutral stimulus in stressed rats, by inducing both active (burying) and avoidant (escape and grooming) coping strategies. Importantly, a neutral stimulus did not trigger stress responses in vehicle-trained rats, suggesting that the effects seen in THC + CBD-trained rats result from the interaction between acute stress and cannabinoid abstinence. Additionally, we found that cannabinoid use induced notable escape behavior in non-stressed rats. Both preclinical and clinical studies support a contention that extensive and prolonged cannabis use is associated with more severe withdrawal symptoms [28–30] and our data indicate that cannabinoid use, regardless of a previous stressful experience, generalizes stress responses to unconditioned stimuli. Since stress generalization is a hallmark of PTSD [1], these preclinical data support longitudinal clinical studies [3, 4] indicating that cannabis use worsens PTSD symptoms. Furthermore, we previously demonstrated that a stress-CS is sufficient to reinstate cocaine, heroin, and alcohol seeking, reflecting an active coping strategy [12, 14]. Accordingly, we found that the stress-CS induced burying behavior in vehicle-trained rats. However, after cannabinoid use, re-exposure to the stress-CS reduced burying while promoting escape and grooming behaviors. Since avoidance behavior is a key criterion for defining PTSD, and avoidance coping strategies are associated with increased activation of the stress axis [31], our findings suggest that cannabis use heightened the salience of the stress-CS.

THC+CBD use alters astroglial plasticity in response to a stress-CS in males

The astroglial coverage of synapses plays a crucial role in synaptic plasticity by supplying metabolic support, gliotransmitters, and maintaining ion homeostasis [16]. Abstinence from cocaine, heroin, and methamphetamine self-administration reduced the proximity of astrocytes and Synapsin-I in NAcore [32–35]. Similarly, we demonstrated that abstinence from THC + CBD reduced the astroglial association with synapses in NAcore, suggesting that the effect is shared by many classes of addictive drugs. Akin to abstinence from addictive drugs, an acute restraint stress also induced enduring decreases in astrocyte-synapse association in NAcore [23]. Consistently, we found that acute stress reduced astrocyte-Synapsin-I association in both vehicle- and THC + CBD-trained rats. Finally, we demonstrated that re-exposure to the stress-CS increased the astrocyte-synaptic association in both vehicle- and THC + CBD-trained rats. However, while the stress-CS only partially restored the astrocyte-synapse association in vehicle-trained rats, cannabinoid exposure potentiated the re-association of astrocytes to synapses induced by the stress-CS. In examining astrocyte-Synapsin-I association related to coping strategies, we found a negative correlation between astrocyte-Synapsin-I colocalization and grooming behavior in vehicle-trained rats. This implies that astroglial-synaptic association may partly countermand the expression of avoidant coping behaviors, although this correlation vanished following THC + CBD usage. Conversely, in THC + CBD-trained males, the astrocyte-Synapsin-I association was negatively correlated with burying behavior. Overall, our data suggest that stress-CS exposure during cannabinoid abstinence alters how astrocyte-synaptic association in the NAcore influences coping strategies.

THC+CBD use following an acute stress reduces Synapsin-I density in NAcore

Severe stressors are known to decrease the levels of synaptic markers, including Synapsin-I and PSD95, in several brain regions. For example, chronic unpredictable mild stress (CUMS) decreases the expression of both Synapsin-I and PSD95 in the prefrontal cortex and hippocampus [24, 36] and a single prolonged stress (SPS) reduces Synapsin-I and PSD95 in the hippocampus [37]. In NAcore, the combination of SPS and foot shock reduces PSD95 [38]. In contrast, we found that neither 2 h of acute restraint stress, stress-CS exposure, nor abstinence from THC + CBD without stress pre-exposure affected the Synapsin-I density in the NAcore [7]. Thus, the reduction seen in astrocyte-synaptic connection produced by stress pre-exposure (stress-NS) or THC + CBD without stress pre-exposure is largely due to astrocytic motility not an overall reduction in synapses. In contrast, we found that both Synapsin-I and astroglial-synaptic association were reduced in the group receiving acute restraint stress and cannabinoid use (Stress-NS). Thus, the retraction of astrocytes induced by cannabinoid abstinence, along with the loss of presynapses resulting from the interplay between stress and cannabinoid use, may contribute to how THC + CBD generalizes stress responses to a neutral stimulus.

THC+CBD use induces astrocyte atrophy during exposure to the stress-CS

It is well-documented that stress alters astrocyte morphology [25]. For example, chronic restraint stress reduced the process length, branching, and volume of astrocytes in the prefrontal cortex [39], and astrocyte volume in the amygdala [40]. CUMS also induced astrocyte atrophy in the prefrontal cortex and hippocampus [41, 42]. Moreover, the SPS protocol led to astrocyte atrophy in the hippocampus [43], where astrocyte atrophy is linked with anxiety and depressive-like behavior. For instance, astrocyte morphology negatively correlates with anxiety, while treatments restoring the astrocyte morphology alleviate symptoms of anxiety and depression in animal models [42–44]. In NAcore, neither acute restraint stress nor the stress-CS affect astroglial morphology [23].

In contrast with stress, the effects of THC on astrocyte morphology are less understood. In organotypic hippocampal slices, prolonged incubation with THC reduces astrocytic processes [45]. Astrocytic processes are also reduced in the amygdala of rats treated with high doses of THC during adolescence [46]. However, the effects of voluntary cannabinoid consumption on astrocyte structure were unknown. Here, we demonstrated that THC + CBD self-administration, with or without a prior stressful experience, did not affect measures of astrocyte morphology, including surface area and volume. However, re-exposure to the stress-CS triggered astrocyte atrophy in THC + CBD, not vehicle-trained rats. Surprisingly, we observed a negative correlation between astrocyte surface area and volume and astrocyte-Synapsin-I colocalization, indicating that cannabinoid use and stress-CS triggered both astrocyte atrophy and increased astroglial association to synapses.

THC+CBD use changes the function of MMP-2,9 in males

The gelatinases MMP-2 and MMP-9 play a necessary role in activity-dependent multipartite synaptic plasticity [16]. For example, MMP-9 controls dendritic spine enlargement induced by long-term potentiation [47]. Thus, over the last 20 years, research has focused on the role of MMP-2,9 in various psychiatric disorders, including drug and stress-related disorders [17]. Our laboratory demonstrated that drug seeking induced by Pavlovian cues requires transient activation of MMP-9 in the NAcore, which leads to both spine head expansion and the insertion of AMPA receptors on D1-medium spiny neurons [22, 48, 49]. In parallel, abstinence from cocaine or heroin self-administration constitutively increases the MMP-2 activity around D2-MSNs [21, 22]. Concerning stress-related disorders, both acute and chronic restraint stress increase MMP-9 activity in the hippocampus [50, 51]. We also found that acute restraint stress increases MMP-9 in NAcore [20]. However, the effect of stress-CS on the activity of MMP-2,9 was unknown. Given that both stress-CS and drug cues promote spine head expansion [11, 48], we hypothesized that stress-CS would increase MMP-2,9 activity. As anticipated, MMP-2,9 activity was increased after exposure to stress-CS in vehicle-trained rats. We also demonstrated that THC + CBD abstinence constitutively increased MMP-2,9 activity. The combination of stress pre-exposure and cannabinoid abstinence (stress-NS) did not alter the basal activity of MMP-2,9. However, re-exposure to the stress-CS during cannabinoid abstinence increased the MMP-2,9 activity. Cannabinoid use not only elevated MMP-2,9 activity but also contributed to a shift from active to avoidant coping behaviors. Thus, while MMP-2,9 activity positively correlated with burying in vehicle-trained rats, cannabinoid use reversed the positive correlation to a negative correlation with burying and revealed a positive correlation with escape behavior.

MMP-2 regulates astrocyte plasticity induced by the stress-CS following cannabinoid use

Considering that stress-CS re-exposure after cannabinoid use strongly activated MMP-2,9 and simultaneously increased astrocyte-synaptic contact and astrocyte atrophy, we hypothesized that these events are interconnected. To establish a causal relationship between MMP-2,9 activity and astroglial plasticity, we administered inhibitors of both gelatinases before the DBT. We found that inhibiting either MMP-2 and MMP-9 reduced astroglial contact with synapses in NAcore, while only the MMP-2 inhibitor reduced Synapsin-I density and prevented astrocyte atrophy. Additionally, inhibiting MMP-2 decreased grooming while increasing burying, indicating that MMP-2 activation by stress-CS may contribute to the switch from active to avoidant behaviors in THC + CBD withdrawn rats. In summary, these data constitute the first in vivo evidence linking the catalytic activities of MMP-2 and MMP-9 to astroglial plasticity and stress-associated behaviors, although they do not demonstrate a causal chain from MMP-2,9 activation to astroglial plasticity to changing the behavioral profile induced by stress-CS.

Sexual dimorphism

Women are twice as likely to develop PTSD and exhibit greater vulnerability for anxiety-related disorders [51–53]. However, historically, preclinical studies on stress-related disorders mostly used males and focused primarily on freezing behavior [52]. Including females and assessing other stress responses introduced new insights into coping strategies [53, 54]. Indeed, research demonstrates that male rats freeze more than females, while females display higher escape-like behaviors [52, 55–57]. Accordingly, we found that females escape more than males across all stress conditions. Surprisingly, neither stress nor THC + CBD had a significant impact on coping strategies in females. This may be due to high variability we observed in stress responses between females. Although the estrous cycle is not involved in escape-like behavior [42], it would be interesting to evaluate its influence on other coping responses. Despite not affecting behaviors, stress and cannabinoid use changed multipartite plasticity in females. Unlike males, the number of synapses (Synapsin-I puncta) is increased in stressed rats following THC + CBD use. Regarding astroglial plasticity, cannabinoid abstinence without stress pre-exposure increased both the surface area and volume of NAcore astrocytes in females compared to males. This result differs from cocaine, where prolonged abstinence elicits NAcore astrocyte atrophy in males without affecting female astrocytes [32]. Furthermore, while acute restraint stress did not affect astrocyte morphology in males, it induced astrocyte atrophy in females. The opposite effect is observed after chronic stress, where CUMS reduces astrocyte morphology in males without affecting females [58]. Our findings contribute to research on sexual dimorphism in astroglial plasticity. More studies are needed to understand how stress and cannabis differently affect female and male multipartite plasticity.

Clinically relevant targets

This study identified potential neurobiological mechanisms underpinning the effects of cannabinoid use in male rats on coping behaviors elicited by a conditioned stress, including MMP-2,9 activation and astroglial plasticity in the NAcore. The possibility that these findings may be relevant to clinical symptoms of stress-induced disorders such as PTSD can be seen in recent studies demonstrating elevated plasma levels of MMP-2 and MMP-9 in PTSD patients [59, 60]. Moreover, the nonselective MMP inhibitor doxycycline disrupts the consolidation of cued fear memory in humans [61]. Additionally, a single dose of minocycline, a more specific MMP-9 inhibitor, administered before a contextual fear memory paradigm, reduced fear memory consolidation [62]. These clinical data suggest that MMP inhibition represents a potential strategy to treat the comorbidity between PTSD and CUD. Another interesting strategy would be to target astroglial plasticity. For example, the antidepressant fluoxetine alleviates depressive-like symptoms, partly by reversing astrocyte atrophy caused by CUMS [42]. Moreover, in humans, antidepressant treatment reverses astroglial atrophy in the hippocampus [63]. Thus, exploring the impact of antidepressant treatment on astrocyte atrophy resulting from stress-CS after cannabis use would be intriguing.

Author contributions

RH designed the study, analyzed the data, and interpreted the results with input from PWK RH performed surgeries. RH, LD, and ED conducted operant behavioral training. RH carried out confocal analyses. RH produced the figures and authored the manuscript with input and revisions from PWK

Funding

Open access funding provided by the Carolinas Consortium.

Data availability

The data supporting the findings of this study are available from the corresponding authors upon request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

All animal experiments and procedures received approval from the Animal Care and Use Committee of the Medical University of South Carolina (IACUC-2019-00872-1) and were conducted following National Institutes of Health guidelines. No human participants were involved in this study.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41380-026-03687-0.