Synaptic Adaptations in the Amygdala During Alprazolam Withdrawal: Insights from a Female Rat Model of Benzodiazepine Withdrawal Authors: Tamara Radukic, Jelena Martinovic, Milorad Dragic, Milica Zeljkovic Jovanovic, Danica Popovic, Nebojsa Jasnic, Andjela Stekic, Marina Zaric Kontic Published: 2026 ABSTRACT Background/Objectives: Benzodiazepine (BZD) withdrawal is frequently associated with altered emotional reactivity and behavior. Given the higher prevalence of anxiety disorders and the more frequent use of BZDs among women, understanding the neurobiology of BZD withdrawal in females is particularly important as it remains elusive. Methods: Adult female Wistar rats received 8 days of chronic, escalating-dose alprazolam (ALP) treatment followed by spontaneous withdrawal assessed at 36 h and 96 h from the last dose. Behavioral assessment was performed using the Open Field Test. Synaptic protein expression was analyzed by Western blot; neurotransmitter and neuromodulator levels were determined using high-performance liquid chromatography (HPLC) and commercial assay kits; and dendritic spine density and morphology were evaluated by Golgi–Cox staining. Results: ALP withdrawal significantly increased levels of anxiety-like behavior while reducing locomotor and exploratory activity. These behavioral changes were accompanied by increased glutamate levels and protein expression of vGluT1, GluR1, NR1, NR2B, and PSD-95, followed by delayed upregulation of EAAT1 and EAAT2. The GABAergic system exhibited heterogeneous remodeling, characterized by increased GAD65, α1-containing GABAA receptors, and gephyrin, and reduced vGAT expression. Dopamine and serotonin levels remained unchanged, whereas D1 receptor levels were transiently reduced during 36 h withdrawal. The discontinuation of ALP was further associated with temporally selective dendritic spine remodeling, characterized by an early, sustained reduction in branched spine density, a transient increase in filopodia at 36 h, and a reduction in total spine density at 96 h. Conclusions: These findings demonstrate that ALP withdrawal leads to coordinated neurochemical, molecular, and structural remodeling within the amygdala, involving both excitatory and inhibitory synaptic compartments. 1. Introduction Benzodiazepines (BZDs) are among the most widely prescribed psychoactive medications and are extensively used to treat anxiety and fear-related disorders, insomnia, epilepsy, and other neurological and psychiatric conditions. However, BZDs are also among the most frequently misused prescription drugs and carry a substantial risk of abuse, tolerance, physical dependence, and development of withdrawal symptoms after dose reduction or treatment discontinuation [1,2,3,4]. Benzodiazepine withdrawal syndrome includes a wide range of psychological, behavioral, cognitive, and physical symptoms such as increased anxiety, hyperarousal, irritability, sleep disturbances, cognitive dysfunction, and affective dysregulation, posing a significant clinical challenge that often complicates treatment discontinuation [2,5,6,7]. Compared to other BZDs, alprazolam (ALP) is one of the most frequently used and most misused [8]. Due to its high potency and relatively short elimination half-life, ALP is associated with an increased risk of dependence and withdrawal-related complications compared with several other BZDs [8,9,10]. Consequently, prolonged ALP exposure provides a useful experimental framework for investigating the neurobiological mechanisms underlying BZD withdrawal. Because the BZD therapeutic effects are mostly mediated through positive allosteric modulation of γ-aminobutyric acid (GABA) type A receptors (GABAARs), the neurobiology of BZD dependence and withdrawal was primarily interpreted through adaptive changes in GABAergic neurotransmission [1,2,11,12]. Chronic BZD exposure induces multiple compensatory alterations, including changes in GABAAR subunit composition, receptor trafficking and coupling, and overall receptor function [2,12]. However, accumulating evidence suggests that withdrawal-associated behavioral and neurobiological abnormalities cannot be fully explained by GABAergic mechanisms alone. Experimental studies have demonstrated that BZD withdrawal is accompanied by increased anxiety-like behavior, enhanced neuronal excitability, and alterations in glutamatergic neurotransmission, supporting the involvement of broader neuroadaptive processes that extend beyond direct GABAergic modulation [1,2,7,10,12,13,14,15]. Among the brain regions involved in emotional regulation and stress responsiveness, the amygdala serves as a key integrative hub that continuously processes sensory, contextual, and interoceptive information, assigns emotional significance to environmental stimuli and shapes behavioral, autonomic, and endocrine responses [16]. Given its central role in coordinating adaptive responses to emotionally salient experiences, the amygdala demonstrates a remarkable capacity for experience-dependent neuroplasticity. Both stress exposure and chronic pharmacological interventions induce structural and molecular remodeling within the amygdala, including changes in neurotransmitter signaling, synaptic organization, and dendritic spine architecture. These adaptations are increasingly recognized as important contributors to persistent changes in emotional reactivity and stress responsiveness [17,18,19,20]. However, despite substantial scientific interest in the neurobiology of BZD withdrawal, the molecular and structural mechanisms underlying withdrawal-induced adaptations within the amygdala remain understudied. An additional limitation of the available literature is its pronounced sex bias. Although women are more frequently affected by anxiety disorders and receive BZD prescriptions more often than men [21,22], the neurobiology of BZD withdrawal has been studied predominantly in males [14,15]. This discrepancy is especially relevant given the well-documented sex differences in brain structure and function, synaptic plasticity, and stress responsiveness [23,24], all of which may influence neuroadaptive responses to chronic drug exposure and withdrawal [25]. Therefore, the current preclinical findings related to females remain neglected. In summary, although behavioral manifestations of BZD withdrawal have been documented, much less is known about the accompanying neuroadaptations within the amygdala, especially in females. Therefore, this study used a translationally relevant ALP withdrawal paradigm in adult female Wistar rats and investigated withdrawal-associated alterations across multiple levels of biological organization. Specifically, we examined withdrawal-associated behavioral alterations together with changes in neurotransmitter/neuromodulator content, synaptic protein expression, and dendritic spine morphology within the amygdala. We hypothesized that ALP withdrawal would be associated with coordinated neurochemical and synaptic remodeling within the amygdala. 2. Results 2.1. Behavioral Alterations Following ALP Withdrawal Prior to behavioral testing, routine neurological and sensorimotor screening revealed no overt abnormalities in neurological, sensory, vestibular, or gross motor functions in the experimental animals; the corresponding data are available in the Zenodo repository (DOI: 10.5281/zenodo.21197979). The potential influence of repeated VEH administration on OFT behavior was first evaluated in a separate pilot analysis by comparing baseline measurements with those obtained from the same animals 36 h after the final VEH injection. No significant differences were observed in total distance traveled (p = 0.3542, Supplementary Figure S1A), average speed (p = 0.3574, Supplementary Figure S1B), number of immobility episodes (p = 0.1106, Supplementary Figure S1C), total time immobile (p = 0.1462, Supplementary Figure S1D), number of entries into the center (p = 0.4881, Supplementary Figure S1E), time spent in the center (p = 0.8830, Supplementary Figure S1F), or time spent in the corners of the arena (p = 0.4200, Supplementary Figure S1G). Thus, repeated VEH administration did not significantly alter locomotor, exploratory, or OFT parameters related to emotional reactivity relative to baseline. Behavioral assessment in the OFT revealed that 36 h of ALP withdrawal significantly altered both locomotor and parameters related to anxiety-like behaviour (Figure 1). Compared with the VEH group, animals subjected to ALP withdrawal showed a significant reduction in total distance traveled (t = 3.902, df = 65, p = 0.0002; Figure 1A) and average speed (t = 3.874, df = 65, p = 0.0003; Figure 2B), indicating reduced locomotor and exploratory activity. Analysis of anxiety-related parameters further demonstrated that ALP withdrawal significantly decreased both the number of entries into the center (t = 3.116, df = 65, p = 0.0027; Figure 1C) and time spent in the center of the arena (t = 2.982, df = 65, p = 0.0040; Figure 2D). In contrast, animals in the withdrawal group spent significantly more time in the corners of the arena than VEH-treated animals (t = 3.847, df = 63, p = 0.0003; Figure 2G). In addition, ALP withdrawal markedly increased both the number of immobility episodes (t = 3.678, df = 65, p = 0.0005; Figure 1E) and the duration of immobility (t = 2.933, df = 65, p = 0.0046; Figure 2F). Representative track and heat plots are shown in Figure 2H. To further distinguish changes in spatial exploration from the concomitant reduction in overall locomotor activity, additional locomotion-adjusted and compartment-specific analyses were performed (Supplementary Figure S2). Distance traveled in the center, expressed as a percentage of the total distance traveled by each animal, was significantly reduced in the 36 h ALP group compared with VEH (p = 0.0384; Supplementary Figure S1E). Absolute distance traveled in the center did not differ significantly between groups (p = 0.1330; Supplementary Figure S1A), whereas average speed in the center was significantly increased in ALP-withdrawn animals (p = 0.0146; Supplementary Figure S1B). In contrast, both distance traveled (p = 0.0004; Supplementary Figure S1C) and average speed (p = 0.0001; Supplementary Figure S1D) in the periphery were significantly reduced compared with VEH controls. Thus, locomotor alterations during 36 h ALP withdrawal were spatially heterogeneous rather than uniformly expressed across the arena. 2.2. Glutamatergic Adaptations in the Amygdaloid Region Following ALP Withdrawal Glutamatergic adaptations associated with ALP withdrawal were assessed by determining changes in principal glutamatergic signaling components and glutamate levels in the amygdaloid region after 36 h of withdrawal and by analyzing the expression of glutamatergic synaptic proteins at 36 h and 96 h of withdrawal (Figure 2). Among presynaptic glutamatergic markers, treatment significantly affected the expression of vesicular glutamate transporter 1 (vGluT1; F(2,18) = 5.77, p = 0.0115, Figure 2A). Tukey’s post hoc analysis showed significantly higher vGluT1 expression in both the 36 h (p = 0.0274) and 96 h (p = 0.0188) withdrawal groups compared with VEH, with no significant difference between the two withdrawal time points. Treatment also significantly affected expression of the GluR1 subunit of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR; F(2,15) = 13.79, p = 0.0004, Figure 2B). GluR1 expression was significantly increased after both 36 h (p = 0.0068) and 96 h (p = 0.0004) of withdrawal compared with the VEH group, without changes between withdrawal time points. Similarly, treatment significantly affected expression of the N-methyl-D-aspartate receptor (NMDAR) NR1 subunit (F(2,18) = 14.83, p = 0.0002, Figure 2C). NR1 expression was increased in both the 36 h (p = 0.0016) and 96 h (p = 0.0002) withdrawal groups compared to VEH, while no changes were observed between withdrawal time points. Interestingly, while NR2A subunit expression did not differ among the experimental groups (F(2,18) = 0.27, p = 0.7647; Figure 2D), NR2B subunit expression showed significant temporal alterations (F(2,18) = 5.72, p = 0.0119, Figure 2E). Post hoc analysis revealed a significant upregulation in the 36 h ALP group compared with VEH (p = 0.0099), whereas, interestingly, neither the comparison between VEH and the 96 h ALP group nor between the two withdrawal groups reached statistical significance. Next, treatment significantly affected expression of the scaffold protein postsynaptic density protein 95 (PSD-95; F(2,18) = 18.09, p < 0.0001, Figure 2F). PSD-95 expression was significantly increased in both the 36 h (p = 0.029) and 96 h (p < 0.0001) ALP groups compared with VEH, with no difference between the two withdrawal time points. Finally, the glutamate transporters, excitatory amino acid transporter 1 and 2 (EAAT1 and EAAT2, respectively), exhibited a similar pattern of expression following ALP withdrawal, with both proteins being significantly upregulated only after 96 h withdrawal. Treatment significantly affected EAAT1 expression (F(2,16) = 10.88, p = 0.0010, Figure 2G). EAAT1 expression was significantly increased in the 96 h ALP group compared with both the VEH (p = 0.0026) and 36 h ALP (p = 0.0033) groups, whereas no significant difference was observed between the VEH and 36 h ALP groups. Likewise, treatment significantly affected EAAT2 expression (F(2,17) = 8.44, p = 0.0028). EAAT2 expression was also significantly increased in the 96 h withdrawal group compared with both the VEH (p = 0.0111) and 36 h withdrawal (p = 0.0036) groups, whereas no significant difference was observed between the VEH and 36 h withdrawal groups. In the end, glutamate concentrations were measured in the amygdaloid region and results revealed a significant increase after 36 h of ALP withdrawal (t = 2.639, df = 14, p = 0.0195; Figure 2I). 2.3. Adaptations in GABAergic Signaling in the Amygdaloid Region Following ALP Withdrawal GABAergic signaling adaptations associated with ALP withdrawal were assessed by analyzing the expression of key presynaptic and postsynaptic GABAergic proteins at 36 h and 96 h of withdrawal and by measuring GABA levels in the amygdala after 36 h of withdrawal (Figure 3). Treatment significantly affected expression of the vesicular GABA transporter (vGAT; F(2,18) = 7.01, p = 0.0056, Figure 3A). vGAT expression was significantly reduced after 36 h of withdrawal compared with the VEH group (p = 0.0142), but returned to control levels after 96 h, resulting in a significant difference between the two withdrawal groups (p = 0.0102) but not between the VEH and 96 h groups. Treatment significantly affected expression of the GABA-synthesizing enzyme glutamate decarboxylase 65 (GAD65; F(2,18) = 4.61, p = 0.0241, Figure 3B). GAD65 expression was significantly increased only in the 96 h withdrawal group (p = 0.0186) compared with VEH animals, with no significant difference between the two withdrawal time points and between VEH and 36 h timepoint. Interestingly, treatment significantly affected expression of the inhibitory postsynaptic scaffold protein gephyrin (F(2,16) = 6.87, p = 0.0070, Figure 3C). Gephyrin expression was significantly increased after both 36 h (p = 0.0097) and 96 h (p = 0.0283) of withdrawal compared with the VEH group, with no significant differences between the two withdrawal time points. Alprazolam withdrawal also significantly affected expression of the α1 subunit of GABAAR (F(2,18) = 7.93, p = 0.0034, Figure 3D). Expression of the α1 subunit was significantly increased in both the 36 h (p = 0.0144) and 96 h withdrawal groups (p = 0.0046) compared with VEH, with no significant difference between the withdrawal time points. Finally, GABA concentration in the amygdala did not differ significantly between the VEH and 36 h ALP groups (t = 1.302, df = 15, p = 0.2125; Figure 3E). 2.4. Dopaminergic and Serotonergic Adaptations in the Amygdaloid Region Following ALP Withdrawal Dopaminergic adaptations associated with ALP withdrawal were evaluated by measuring the expression of dopamine D1 and D2 receptors (D1R and D2R) at both 36 h and 96 h of withdrawal, as well as dopamine levels in the amygdaloid region after 36 h of withdrawal. Serotonergic adaptations were assessed solely by measuring serotonin levels in the amygdaloid region at the 36 h withdrawal time point (Figure 4). Treatment significantly affected dopamine D1 receptor (D1R) expression (F(2,18) = 21.01, p < 0.0001), whose expression was significantly reduced after 36 h of withdrawal compared with the VEH group (p < 0.0001). At 96 h of withdrawal, D1R expression was significantly increased compared with 36 h (p = 0.0013) but remained significantly lower than in the VEH group (p = 0.0260). In contrast, dopamine D2 receptor (D2R) expression did not differ among the experimental groups (F(2,18) = 0.64, p = 0.5351; Figure 4B). Neither dopamine (t = 0.9949, df = 12, p = 0.3394; Figure 4C) nor serotonin (t = 1.438, df = 12, p = 0.1759; Figure 4D) concentrations differed significantly between the VEH and 36 h ALP withdrawal groups. 2.5. Structural Synaptic Adaptations in the Basolateral Amygdala Following ALP Withdrawal Structural synaptic adaptations associated with ALP withdrawal were evaluated by analyzing dendritic morphology, dendritic spine density and subtype distribution and synaptophysin expression in the BLA. Synaptophysin (SYN) expression was first evaluated as a marker of presynaptic terminals. Treatment significantly affected synaptophysin expression (F(2,18) = 6.5, p = 0.0075, Figure 5A). Synaptophysin expression was significantly increased after 96 h of withdrawal compared with the VEH group (p = 0.0065), whereas no significant differences were observed between the VEH and 36 h withdrawal groups or between the two withdrawal time points (Figure 5A). To exclude differences in dendritic sampling as a potential confounding factor, the lengths of analyzed dendritic segments were first compared and found not to differ among the experimental groups (F(2,173) = 0.50, p = 0.6014; data are available in the Zenodo repository; DOI: 10.5281/zenodo.21197979). Post hoc analysis likewise confirmed that no pairwise differences reached statistical significance. Analysis of total dendritic spine density revealed a significant effect of withdrawal (F(2,9) = 4.972, p = 0.0351; Figure 5B). Total spine density showed a progressive decrease throughout the withdrawal, reaching a significant reduction at 96 h compared with the VEH group (p = 0.0402), whereas no significant difference was observed at 36 h (Figure 5B). To identify which dendritic spine populations contributed to the reduction in total spine density of BLA, individual spine subtypes were subsequently analyzed. Branched spine density was significantly decreased (F(2,8) = 8.834, p = 0.0094; Figure 5I) in both withdrawal groups relative to VEH (p = 0.0224 and p = 0.0103, respectively), with no significant difference between the 36 h and 96 h withdrawal groups. Stubby spine density did not show a significant overall effect of withdrawal (F(2,9) = 3.962, p = 0.0583, Figure 5I); however, post hoc analysis revealed a significant reduction after 36 h (p = 0.0497) compared with the VEH group, while no significant differences were observed between the two withdrawal groups or between the 96 h withdrawal to VEH group. Filopodia spine density showed a significant effect of withdrawal (F(2,9) = 4.364, p = 0.0473; Figure 5I), with a significant increase at 36 h compared with the VEH group (p = 0.0413), whereas no significant difference was observed between the two withdrawal groups or 96 h compared to the VEH group. In contrast, mushroom (F(2,9) = 1.101, p = 0.3734, Figure 5I), thin (F(2,9) = 1.878, p = 0.2081, Figure 5I) and long-thin spine densities (F(2,9) = 0.07933, p = 0.9244, Figure 5I) did not differ significantly among the experimental groups. Representative BLA pyramidal neurons and corresponding dendritic segments from the VEH, 36 h ALP, and 96 h ALP groups are shown in Figure 5C–H. 3. Discussion The present study provides a comprehensive characterization of ALP withdrawal in adult female rats by integrating behavioral assessment with neurochemical, synaptic, and dendritic spine analyses. Whereas previous experimental studies have largely focused on individual aspects of BZD withdrawal, such as behavioral manifestations or receptor adaptations, the present study examined complementary levels of organization within the same experimental framework. This integrative approach offers a broader understanding of the neurobiological processes underlying BZD withdrawal. A principal finding of the study is that 36 h ALP withdrawal produced a highly consistent behavioral phenotype in the OFT. Rather than relying on a single behavioral parameter, withdrawal was characterized by reduced locomotor and exploratory activity, decreased center exploration, increased corner occupancy, and prolonged immobility, with all measured parameters indicating the same behavioral outcome. The consistency of these alterations across a relatively large experimental cohort further strengthens the robustness of the observed withdrawal phenotype. These findings are consistent with previous studies demonstrating increased anxiety-like behavior following withdrawal after chronic treatment with different BZDs, despite substantial differences in treatment duration, dosing regimens, and behavioral paradigms [26,27,28,29,30,31]. Nevertheless, the observed reduced locomotor activity during withdrawal requires careful interpretation because it could theoretically reflect residual sedative drug effects rather than withdrawal itself. Importantly, the additional locomotion-adjusted, compartment-specific analyses provide a more nuanced interpretation of the behavioral phenotype observed during early ALP withdrawal. Although overall locomotor activity was reduced, this effect was not uniformly expressed across the OFT arena. Both distance traveled and average speed were markedly reduced in the periphery, whereas absolute distance traveled in the center remained unchanged. Moreover, when distance traveled in the center was normalized to the total distance traveled by each animal, the relative proportion of locomotor activity occurring within the center remained significantly reduced. This finding indicates that the altered pattern of center exploration cannot be attributed solely to the overall reduction in locomotor activity. Interestingly, ALP-withdrawn animals exhibited increased movement speed specifically within the center despite reduced locomotor activity in the periphery. Given the exposed and potentially anxiogenic nature of the OFT center, this spatially selective increase in speed may reflect more rapid traversal of the center rather than sustained exploration of this area. Taken together with the reduced proportion of total locomotion occurring within the center, these findings suggest an altered spatial organization of exploratory behavior during early withdrawal rather than simple generalized locomotor suppression. Nevertheless, because movement speed alone does not directly measure escape or avoidance motivation, this interpretation should be considered cautiously. Additional preliminary, unpublished analyses performed during revision further argue against this possibility. Using a validated gas chromatography-mass spectrometry method, parent ALP was not detected in hippocampal tissue at either 36 h or 96 h after the final administration (