Sexually dimorphic cannabinoid 1 receptors on CaMKIIα neurons in the medial prefrontal cortex mediate sex differences in ACEA-induced antinociception in mice

Supplemental Digital Content is Available in the Text.

CB1 receptors on CaMKIIα neurons in the medial prefrontal cortex regulate sex differences in cannabinoid-induced analgesia, with distinct synaptic effects observed between male and female mice.

Abstract

Patients suffer greatly from neuropathic pain, and pharmacological therapies usually prove insufficient. Although the causes are yet unknown, sex differences in reaction to analgesics help to explain poor outcomes. Cannabinoid-based treatments are more effective in female patients, although the reasons are not well understood. The ventrolateral periaqueductal gray (vlPAG) is a crucial node in the descending pain inhibitory system. Our previous research shows that the GABA-vGlut2 inhibitory circuitry in the vlPAG regulates vGlut2 neuron activity in a sex-specific manner to modulate cannabinoid analgesia. However, it remains unknown whether this circuitry is regulated by upstream neurons. This study identified CaMKIIα neurons in the upstream medial prefrontal cortex (mPFC) that control vlPAG^GABA. Deleting cannabinoid receptor 1 (CB₁R) in this pathway causes severe pain in female mice and diminishes vlPAG^vGlut2 inhibition, leading to overall output. These results clarify CB₁R’s function in the mPFC^CaMKIIα-vlPAG^GABA-vlPAG^vGlut2 circuit, shedding light on sex-specific cannabinoid analgesia and providing a basis for developing more potent analgesics for both sexes.

1. Introduction

Globally, 10% of patients suffer from neuropathic pain (NP), with mechanical allodynia as a major contributor to disability. ^7,12 The complex etiology and variable treatment responses of NP present significant clinical challenges, with biological sex emerging as a key factor influencing analgesic efficacy. ^22,31,34,36 The NIH's 2019 report on sex differences in health and IASP's 2024 Global Year on Sex/Gender Disparities in Pain underscore the imperative to integrate sex considerations into pain research.^5,6,23 However, mechanisms underlying sex-specific analgesic responses remain elusive.

G protein-coupled receptors (GPCRs) represent critical targets for pain management, with growing evidence of sex-dimorphic signaling pathways. ^33,39 Cannabinoid-based therapies engaging the endocannabinoid system (ECS) show promise for chronic pain, exhibiting mechanisms distinct from opioids.^3,9 Paradoxically, while female mice frequently exhibit heightened sensitivity across preclinical models, they often demonstrated superior analgesic responses to cannabinoids compared with male mice. This divergence between basal sensitivity and pharmacodynamic efficacy underscores the complexity of sex-specific pain modulation. Pain processing involves ascending transmission and descending modulation pathways. Ascending nociceptive pathways transmit pain signals from peripheral tissues to the brain, and descending inhibitory pathways modulate these signals. Synaptic plasticity at excitatory/inhibitory synapses in these pathways drives central sensitization and mechanical allodynia.²⁸

The ECS modulates pain through presynaptic cannabinoid receptor 1 (CB₁R), which regulates synaptic plasticity and neurotransmitter release.^9,10 CB₁R activation concurrently inhibits ascending nociception and potentiates descending inhibition, yielding net analgesia. ²⁹ Crucially, CB₁R expression exhibits sex-, region-, and cell type-dependent heterogeneity. ^26,53 Crucially, in the descending pain pathway is the midbrain periaqueductal gray (PAG), which sends inhibitory projections to the rostral ventromedial medulla (RVM) and then to the spinal dorsal horn.⁴² The ventrolateral subdivision (vlPAG) is particularly important for cannabinoid-induced analgesia, which is primarily mediated by CB₁R on local GABAergic neurons. ^3,21,51 Sex-specific CB₁R expression on vlPAG^GABA (GAD2-expressing) neurons underlies differential synaptic plasticity and vesicular glutamate transporter 2 (vGlut2)-positive excitatory output neurons activation in female mice.¹⁷ However, it remains unclear how NP alters this circuit or whether upstream CB₁R contributes to its sex-specific dimorphism. The vlPAG receives dense projections from limbic forebrain regions, including the medial prefrontal cortex (mPFC) and hypothalamus, establishing the mPFC-vlPAG circuit as a key player in descending pain control.^16,51 Although emerging evidence implicates endocannabinoids in the mPFC during chronic pain,^43,47 the specific mechanisms, particularly those involving sex-specific CB₁R expression, remain unresolved.

To explain the mechanisms behind sex-specific effects of cannabinoids, we examined whether CB₁R on CaMKIIα-expressing excitatory neurons in the mPFC contributes to differential pain modulation in male and female mice. We hypothesized that CB₁R expressed on mPFC neurons projecting to the vlPAG mediates sex-dependent antinociceptive effects. To evaluate this, we utilized viral tracing, immunohistochemistry, fluorescent in situ hybridization, and RNAscope techniques to map CB₁R expression and identify potential sex differences. We further investigated the functional significance of CB₁R on mPFC^CaMKIIα neurons in neuropathic pain regulation through conditional knockout, opto-XR, and electrophysiology approaches.

2. Methods

2.1. Animals

Male and female C57BL/6J (8-12 weeks old) wild-type mice (Beijing Vital River Laboratory Animal Technology Co., Ltd, China) were used for behavioral, immunohistochemical, and electrophysiological experiments. To generate GAD2-Cre-tdTomato mice, adult homozygous GAD2-Cre (Jackson Laboratories) mice were crossed with Ai9 (Rosa-CAG-LSL-tdTomato-WPRE) reporter mice (Jackson Laboratories). CB₁R-floxed⁴⁶ was constructed by Beijing Biocytogen. CB₁R-iCre-EGFP was obtained from the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences.⁵³ Those weighing 22 to 28 g at 8 to 12 weeks old were chosen. Mice were group housed in a temperature-controlled environment (22-24°C) under a 12-hour light–dark cycle (lights on from 7:00 am to 7:00 pm) and fed an ad libitum diet of chow and water. For the care and use of laboratory animals, the Fourth Military Medical University guidelines were followed. All procedures were approved by the Fourth Military Medical University Ethics Committee (Xi'an, China, ethics approval reference number: IACUC-20230950). Mice were randomly assigned to experimental groups using a random number generator. All behavioral testing and data analysis were performed by experimenters blinded to the genotype and treatment conditions.

2.2. Chronic constriction injury model

In this experiment, a chronic constriction injury (CCI) model of the sciatic nerve was used to study neuropathic pain, following the method described by Bennett and Xie. ² The procedure was summarized as follows: Surgical instruments were routinely sterilized, and mice were anesthetized with 3% isoflurane. The left hind limb of the mouse was exposed, and the area was shaved and disinfected using an electric clipper. A 1.5-cm incision was made in the skin using ophthalmic scissors, and the deep muscles were bluntly separated layer by layer to expose the sciatic nerve. A small, curved hook made from a glass electrode tube was gently used to lift the sciatic nerve. The sciatic nerve was then ligated with 6-0 chromic gut sutures (presoaked in 0.9% saline for 1 day) around the main trunk of the nerve. The appropriate tightness of the ligature was indicated by slight twitching of the left hind limb or minor muscle contractions around the left hind limb. For the sham surgery group, the sciatic nerve was not ligated, but all other surgical procedures were the same as for the CCI group. After the surgery, the incision is sutured, disinfected, and the mouse was placed in a 37°C incubator for observation until it recovered consciousness (able to move independently), after which it was returned to the animal housing for further observation.

2.3. Paw withdrawal mechanical threshold test

Mice were gently placed into a brown behavioral testing box with a fine mesh bottom to conduct the von Frey filament test using the up-down method as described by Chaplan et al. (1994).⁴ Following our previous research,¹⁸ the experimental procedure is briefly outlined as follows: von Frey filaments (North coast) ranging from 0.008 to 2.0 g were selected and applied vertically in an “S” shape to the central plantar surface of the mouse's hind paw. Each stimulus lasted for approximately 5 seconds. A positive response was recorded if the mouse quickly withdrew its hind paw. Each mouse was tested 10 times, with intervals of at least 3 minutes between each trial. When the withdrawal response rate exceeded 50%, the corresponding von Frey filament weight was recorded as the paw withdrawal mechanical threshold.

2.4. Dynamic score test

A fine brush was used to gently stroke the toes of the mice to assess dynamic touch-evoked pain.⁸ Each mouse was tested in a quiet room, placed individually in transparent chambers on an elevated wire mesh platform, and habituated for at least 30 minutes before testing. For mice not subjected to the CCI (Chronic Constriction Injury) model, the measured value represented light touch sensitivity, typically resulting in no paw withdrawal, and was scored as 0. A quick elevation of the stimulated paw or very rapid movement indicated a typical response to mechanical stimulation and was scored as 1. Post-CCI surgery, the measurement focused on dynamic mechanical allodynia. A quick elevation of the stimulated toe or very rapid movement was scored as 0. Sustained elevation of the stimulated toe for more than 2 seconds was scored as 1. A vigorous elevation of the stimulated toe above the mouse's body plane was scored as 2. Licking of the stimulated paw or a rapid withdrawal movement was scored as 3. Each session consisted of 5 trials, with 10-second intervals between strokes within a session. The entire session was then repeated 3 times at ≥3-minute intervals to prevent sensitization. The average score from the 5 repeated sessions was calculated as the final brush score for each individual mouse.

2.5. Paw withdrawal thermal latency—Hargreaves test

The Hargreaves method¹⁴ helped to evaluate thermal nociception. Before testing, mice were separately placed in transparent Plexiglas chambers on an elevated glass platform and allowed to adapt for at least 30 minutes before testing. A focused beam of radiant heat was applied to the plantar surface of the hind paw using a plantar test apparatus (eg, IITC Life Science), and the latency to paw withdrawal was recorded as the paw withdrawal thermal latency (PWTL). The intensity of the heat stimulus was preadjusted so that baseline withdrawal latencies in control animals ranged between 8 and 12 seconds. Each mouse was tested 5 times, with a 10-minute interval between each trial to prevent sensitization. The mean latency of these 5 trials was calculated as the final PWTL for each animal. A cut-off time of 30 seconds was imposed to prevent potential tissue damage.

2.6. Rotarod test

The sensorimotor coordination was detected by an accelerating rotarod system (ZS-YLS-4C, ZS Dichuang, China). First accustomed to the device over 2 consecutive days, mice underwent pretraining sessions including 5-minute trials at a steady speed of 5 rpm. Should a mouse drop during training, the trial was immediately resumed to provide enough acclimatization. Mice underwent an acceleration treatment on test day, whereby their rotarod speed grew linearly from 4 rpm to 40 rpm over a 5-minute period. Every trial yielded a recorded delay to fall. Every mouse finished 2 test runs with an intertrial interval of at least 20 minutes; the mean latency to fall was computed and used as the final performance score.

2.7. Immunohistochemistry

Male- and female-WT mice aged 6 to 8 weeks were selected. The mice were deeply anesthetized with an intraperitoneal injection of 0.5% sodium pentobarbital. Cardiac perfusion was performed first with prechilled 0.9% saline to remove the blood, followed by 40 mL of 4% paraformaldehyde solution (prechilled). The whole brain was removed and fixed in 4% paraformaldehyde (prechilled) for 2 to 3 hours. The brain tissue was dehydrated in 20% sucrose solution in a 4°C refrigerator until the tissue sank and then transferred to 30% sucrose solution for further dehydration (at least 1 day at 4°C). Once the brain tissue had fully sunk, it was removed and the surface was blotted with filter paper to remove excess moisture. The excess tissue was trimmed, the brain was placed on a mold, embedded in embedding medium, and frozen. Moreover, 30-μm-thick frozen sections were cut using a cryostat (Leica, Germany, RRID:SCR_018061). The prepared brain sections were placed in filtered PBS and rewarmed at room temperature for 30 minutes. They were washed with PBS 3 times for 10 minutes each. The sections were incubated in primary antibody dilution containing GABA (GTX125988, 1:200, Rabbit, Genetex), CaMKIIα (#50049, 1:200, mouse, CST), CB₁R (Af450, 1:200, goat, Frontier Institute, Japan, RRID: AB_2571592), and NeuN (266004, 1:400, guinea pig, synaptic systems, USA) at 4°C with shaking for 30 to 36 hours. They were washed with PBS 3 times for 10 minutes each. The sections were incubated in a secondary antibody (1:500, Jackson ImmunoResearch Laboratories) mixture containing Alexa Fluor 594 (donkey anti-mouse), Alexa Fluor 647 (donkey anti-guinea pig), and Alexa Fluor 488 (donkey anti-goat) for 2 hours. They were washed with PBS 3 times for 10 minutes each. The brain sections were mounted onto slides, covered with antifade mounting medium, and coverslips were added for imaging.

2.8. Fluorescent in situ hybridization

CB₁R-iCre-EGFP mice were anesthetized with sodium pentobarbital (0.4%, i.p.) and perfused with DEPC-treated PBS followed by DEPC-treated paraformaldehyde (prechilled). The whole brain was fixed in DEPC-paraformaldehyde at 4°C overnight. The brain was dehydrated in 20% and 30% DEPC-PBS sucrose solutions, and 30 μm frozen sections were prepared. vGlut1 and vGlut2 probes were constructed using primers (vGlut2: forward primer was “CCAAATCTTACGGTGCTACCTC’’ and the reverse primer was ‘‘TAGCCATCTTTCCTGTTCCACT”; vGlut1: forward primer was “CAGAGCCGGAGGAGATGA” and reverse primer was “TTCCCTCAGAAACGCTGG”).²⁵ Sections were washed in DEPC-PBS, rinsed in DEPC-PBS with 0.02% Triton X-100, followed by DEPC-PBS. They were soaked in 0.5% hydrogen peroxide, treated with proteinase K (10 μg/mL) and glycine (0.2%), and acetylated in triethanolamine (1.325%) with 0.25% acetic anhydride. The sections were rinsed with DEPC-PBS. Hybridization buffer (0.2 μg/mL) was prepared, and the digoxigenin-UTP-labeled RNA probe was added and pre-denatured at 100°C for 5 minutes. The brain slices were then incubated with the probe at 56°C for 16 to 18 hours. The sections were washed in ×2 SSC and ×0.2 SSC, rinsed in PBST, blocked with PBST containing 10% donkey serum, and incubated with goat anti-digoxigenin-peroxidase antibody at 37°C for 1 hour. The sections were washed in PBST, incubated with fluorescein amplification mixture, and rinsed with PBST. The sections were incubated overnight at 4°C with RFP antibody (1:500), washed, and incubated with donkey anti-rabbit 594 secondary antibody at room temperature for 2 hours. The sections were rinsed in PBS, counterstained with DAPI (1:5000) for 10 minutes, and mounted and imaged using a slide scanner and confocal microscope for detailed imaging.

2.9. RNAscope in situ hybridization

The RNAscope in situ hybridization assay was performed according to the instructions of the manufacturer (Advanced Cell Diagnostics, ACD). Mice were anesthetized by intraperitoneal injection of pentobarbital sodium and perfused with 4% paraformaldehyde (PFA), followed by postfixation for 24 hours. Brains were dehydrated (20% and 30% sucrose). Subsequently, transverse sections (20-µm-thick) were prepared using a cryostat microtome (Leica, Germany, RRID:SCR_018061) and then baked at 60°C for 30 minutes. The sections were processed with target retrieval reagent and incubated with chicken anti-GFP antibody (Cat # 600-901-215, 1:1000, Rockland) and rabbit anti-RFP antibody (Cat # 600-401-379S, Rockland) overnight at 4°C, followed by washing. Protease Plus was applied to enhance probe binding, and the CB₁R-c1 probe (420721, ACD) or FLPO-probe (520791-C3, ACD) was incubated at 40°C for 2 hours. Amplification steps (AMP1, AMP2, AMP3) were followed by HRP and Opal 690 fluorescent labeling (FP1497001KT, akoya). After labeling, the slides were incubated with Alexa Fluor 488-conjugated donkey anti-chicken (Jackson ImmunoResearch, Cat # 703-546-155) and 594-conjugated donkey anti-rabbit (Cat # 711-585-152; RRID: AB_2340621) and then washed. Finally, slides were mounted with DAPI, and images were captured using a confocal (FV1200, Japan). Negative (310043, ACD) and positive (320881, ACD) probes were used to test the quality of the brain slice (see Fig. S1, supplemental digital content, http://links.lww.com/PAIN/C418). To assess the CB₁R knockout and knockdown rates, we utilized the RNAscope semiquantitative ACD scoring method.⁵² The scoring criteria were defined as follows: 0 for no staining or less than 1 dot per 10 cells; 1 for 1 to 3 dots per cell; 2 indicates 4 to 9 dots per cell with none or very few dot clusters; 3 for 10 to 15 dots per cell and/or less than 10% of dots are in clusters; and 4 for more than 15 dots per cell and/or over 10% of dots in clusters. Histo score (H-score) was calculated according to the equation: H-score = Σ (ACD score × percentage of cells per bin).

2.10. Virus stereotactic injection

Surgical instruments were sterilized, and supplies were prepared. The stereotaxic apparatus was calibrated, ensuring that the X, Y, and Z axes were aligned and the ear bars were level (Stoelting). Furthermore, 8-week-old C57BL/6 mice were anesthetized with 0.4% sodium pentobarbital and placed on the operating table. The incisors and ear bars were secured, and the head and neck were disinfected with iodine tincture. A 1-cm midline incision was made using ophthalmic scissors to expose the skull. The skull surface was exposed, 0.3% hydrogen peroxide was applied, excess tissue was trimmed, and the Bregma and Lambda points were aligned with a maximum deviation of 0.03 mm. A small hole was drilled in the skull, the bone membrane was retracted, and 300 nL of virus was injected using a microinjection pump set at 45 nL/minute. After injection, the needle was left in place for 20 minutes before being withdrawn. The incision was disinfected and sutured. The stereotaxic coordinates for the mPFC were +1.75 mm anteroposterior (AP), ±0.4 mm mediolateral (ML), and −2.45 mm dorsoventral (DV); coordinates for vlPAG were (AP = −4.64 mm, ML = ±0.56 mm, DV = −2.70 mm).

For ablation of CB₁R on the CaMKIIα neurons of mPFC, rAAV-CaMKIIα-Cre-mCherry or rAAV-CaMKIIα-mCherry (viral: 1 × 10¹² particles ml⁻¹, BrainVTA Co., Ltd., China) virus was injected into the mPFC. Three weeks later, the mice were used for pain behavior testing. For the ablation of CB₁R on projection neurons innervating the vlPAG region, rAAV2/R-hSyn-FLP-EGFP or rAAV2/R-hSyn-CaMKIIα-FLP and rAAV2/9-hSyn-fDIO-Cre-mCherry viruses (viral: 1 × 10¹² particles ml⁻¹, BrainVTA Co., Ltd., China) were injected into the vlPAG and mPFC of WT or CB₁R-flox mice, respectively, in both sexes.

For labeling glutamatergic neurons in the vlPAG, the rAAV2/9-CaMKIIα-EGFP virus was injected into the vlPAG. Three weeks later, the mice were used for electrophysiology. For labeling neurons innervating the vlPAG, rAAV2/R-hSyn-FLP-EGFP or rAAV2/9-hSyn-fDIO-Cre-mCherry viruses (viral: 1 × 10¹² particles ml⁻¹, BrainVTA Co., Ltd., China) were injected into the vlPAG and mPFC of WT mice, respectively, in both sexes.

For cell-type-specific retrograde transsynaptic tracing, an equal volume of Cre-dependent rAAV-DIO-EGFP-TVA and rAAV-DIO-RG (viral: 1 × 10¹² particles ml⁻¹, BrainVTA Co., Ltd., China) was mixed and then stereotaxically injected into the vlPAG of GAD2-Cre mice. This allowed for the selective expression of EGFP-TVA and RG in GAD2-positive neurons. After a 3-week recovery period to allow for AAV expression, 300 nL of RV-EnvA-ΔG-dsRed was injected into the same site under biosafety level-2 conditions. Following a 1-week period for rabies virus infection and transsynaptic spread, the animals were killed.

For specific activation of CB₁R on the mPFC^CaMKIIα neurons of WT mice in both sexes, pAAV-CaMKIIα-optoCB₁R-EGFP (viral: 1.48 × 10¹³ particles ml⁻¹, Obio Technology, China) or pAAV-CaMKIIα-EGFP virus (viral: 3.47 × 10¹³ particles ml⁻¹, Obio Technology, China) was injected into the mPFC of WT mice in both sexes.

2.11. Optogenetics

Two weeks after the opCB₁R viral injection into the mPFC, under anesthesia, an optical fiber (core diameter Ø200 μm, numerical aperture 0.37, length 3.5 mm, ThinkerTech Nanjing Bioscience Inc, China) was implanted 0.2 mm above the injection site. After allowing the mice to recover for 1 week, the laser source was adjusted to the appropriate intensity and connected to the implanted optical fiber through an optical fiber cable before performing optogenetic stimulation. Before light exposure, baseline von Frey and Brush responses were tested. Then, a blue light exposure (473 nm, 20 Hz with 20 ms, 2 minutes, ThinkerTech Nanjing Bioscience Inc, China) was delivered through the cable, and pain behavioral changes were recorded at 2 minutes, 20 minutes, 30 minutes, 1 hour, 2 hours, and 3 hours after blue light stimulation. To ensure reproducibility and minimize variability, the behavioral tests were conducted in 4 separate batches, with 2 mice tested per session, resulting in a total cohort of 8 mice. This approach allowed for careful handling and consistent testing conditions across all animals.

2.12. Slice preparation and electrophysiological recording

After pre-oxygenating the sucrose solution (mM: sucrose, 234; KCl, 3.6; CaCl₂, 2.5; MgCl₂, 1.2; NaH₂PO₄, 1.2; NaHCO₃, 25; D-glucose, 12; equilibrated with 95% O₂ + 5% CO₂; 310 to 320 mOsm) for 15 minutes, it was frozen in a −80°C freezer. Simultaneously, the perfusion solution (mM: NaCl, 125; KCl, 2.5; CaCl₂, 2; MgCl₂, 1, NaH₂PO₄, 1.25; NaHCO₃, 26; D-glucose, 25, 310-320 mOsm) was pre-oxygenated in a 35°C water bath. The slicer's workbench was prefilled with ice. Once prepared, 3- to 5-week-old GAD2-Cre-tdTomato mice or WT mice injected with rAAV-CaMKIIα-opCB₁R-mCherry and rAAV-CaMKIIα-EGFP viruses in the mPFC and vlPAG, respectively, were anesthetized with 0.5% sodium pentobarbital. The mice were then decapitated, and their brains were quickly extracted and placed in the slicing solution. The tissue was trimmed and secured on the agar and slicer workbench. Using automatic mode, 4 continuous 300-μm-thick slices of the vlPAG were cut from rostral to caudal. The slices were transferred to the perfusion solution and incubated for 40 minutes.

The isolated brain slice was placed in the recording chamber and secured with a nylon net. Using an infrared upright microscope, the vlPAG region was selected as the recording area, and the cells were observed under a ×40 water lens. The fluorescence illuminator was activated, and tdTomato or EGFP-labeled cells were selected. Healthy cells appeared spindle-shaped and elastic. The target cell was moved to the center of the field, then the water lens was raised to make space for the glass electrode. A P97 electrode puller was used to make the glass electrode, which was filled one-third with electrode internal solution to maintain an input resistance of 5 to 10 MΩ. The MultiClamp 700B was set to voltage clamp mode, and the MPC-200 micromanipulator was used to move the glass electrode close to the tissue. Positive pressure of 0.1 mL was applied with a syringe when near the tissue surface. The positive pressure was removed, and negative pressure was applied to form a high-resistance seal (up to 1 GΩ) at the electrode tip, then the cell membrane was broken.

For recording GAD2-tdTomato cells in electrophysiology combined with optogenetics, the vlPAG region was located, and the tdTomato-expressing GAD2+ neurons were clamped at −70 mV and 0 mV. Blue light stimulation (473 nm, 5 mW) was delivered near the EGFP-expressing fiber terminals to record oEPSCs or oIPSCs as controls. Amplitude changes of oEPSCs or oIPSCs were recorded during the perfusion of 10 μM ACEA and the washout phase.

Similarly, when recording CaMKIIα-EGFP fluorescent cells, the cells were clamped at −70 mV and 0 mV. Blue light stimulation (473 nm, 5 mW) was applied near the mCherry-expressing fiber terminals, and sEPSCs or sIPSCs were recorded as controls. Changes in the frequency or amplitude of sEPSCs or sIPSCs were recorded after light stimulation. Signals were acquired using an Axopatch 700B amplifier (Molecular Devices), digitized at 10 kHz with a digitizer (Digidata 1550A, Molecular Devices), and analyzed with pClamp10.0 software (Molecular Devices).

2.13. Statistics

All imaging data of mPFC regions were analyzed based on 3 to 5 or 12 sections per mouse, with a total of 3 to 6 mice per sex group (n = 3-6 mice/group). Data were analyzed using Imaris (Bitplane AG, Switzerland), GraphPad Prism 8.3 (GraphPad Software Inc.), and IBM SPSS 26.0. All data are expressed as mean ± SEM. The normality of data distribution was assessed using the Shapiro-Wilk test before applying parametric tests. Unpaired and paired Student t tests were used to analyze single-variable differences. For static outcomes (H-score/frequency/amplitude), 2-way ANOVA (factors: sex × treatment) evaluated main/interaction effects, with sex-stratified group comparisons performed regardless of interaction significance; for longitudinal data, 3-way repeated measures ANOVA (sex × group × time) using Greenhouse-Geisser correction for sphericity violations was conducted, followed by prespecified exploratory analyses irrespective of global interaction results: (1) sex-stratified group comparisons across time, (2) pairwise group comparisons within each sex-time combination, and (3) detection of timepoint-specific sex × treatment interactions using 2-way ANOVA with Benjamini-Hochberg false discovery rate (FDR) correction. Chi-square test or Fisher exact test was applied to assess proportional differences between groups. P values were deemed statistically significant if P < 0.05. No test for outliers was conducted.

3. Results

3.1. The neurons from the medial prefrontal cortex projecting to the ventrolateral periaqueductal gray region are predominantly CaMKIIα excitatory neurons

The mPFC receives and integrates information from multiple brain regions and then transmits this processed information to nearly all subcortical areas. It plays a crucial regulatory role in the entire pain modulation network.⁴⁸ To further validate the connections between vlPAG and mPFC, we separately injected rAAV2/R-hSyn-FLP-EGFP and rAAV-hSyn-fDIO-mCherry virus into the vlPAG and mPFC regions (Fig. 1A). Immunohistochemistry conducted on the 21st day postinjection confirmed that neurons expressing FLP-EGFP universally co-expressed mCherry, highlighting a robust connection from mPFC to vlPAG (Fig. 1B). Given that this study primarily investigates whether the vlPAG^GABA, as the core of feedforward inhibitory circuit of descending inhibitory system, is regulated by sex-specific mechanisms in the upstream mPFC region, we proceeded by injecting a helper mixture of viruses (rAAV-EF1α-DIO-H2B-EGFP-T2A-TVA-WPRE-hGH-pA and rAAV-EF1α-DIO-RVG-WPRE-hGH-pA in a 1:1 ratio) into the vlPAG of GAD2-Cre (Figs. 1C–E). After 3 weeks, RV-ENVA-△G-dsRed was injected at the same site, followed by immunohistochemistry after 1 week. In situ observations in vlPAG revealed neurons co-labeled with the helper viruses and RV virus (Fig. 1F), confirming successful virus expression. As depicted in Figure 1G, neurons retrogradely traced with DsRed labeling were observed in the mPFC regions of both female and male mice with no sex difference (Fig. 1G, Table 1).

Figure 1.

There are specific projections from the mPFC to the vlPAG. (A) Diagram of rAAV2/R-hSyn-FLP-EGFP or rAAV2/9-hSyn-fDIO-Cre-mCherry virus injections into vlPAG and mPFC of WT mice. (B) Co-labeling of mCherry and EGFP in the mPFC area of WT mice; quantification was based on 3 to 5 sections per mouse, n = 3 mice. Scale bar: 100 μm. (C) Timeline schematic of the helper and rabies virus injection. (D) Structure of the helper and rabies viruses. (E) Schematic diagram of the injection sites for the helper and rabies viruses. (F) Representative immunofluorescence images of the triple virus in the vlPAG. The image on the right is an enlarged view of the dashed area on the left. Arrows indicate neurons co-labeled by the helper and rabies viruses. Analysis was based on 3 to 5 sections per mouse, with n = 3 male and 3 female mice. (G) Brain regions containing upstream neurons of GABAergic neurons in the vlPAG. cg1, cingulate cortex 1; cg2, cingulate cortex 2; LSI, lateral septal nucleus, intermediate part; Acb, accumbens nucleus; M2, secondary motor cortex; M1, primary motor cortex; S1HL, primary somatosensory cortex, hindlimb region; LH, lateral hypothalamic area; CeM, central amygdaloid nucleus, medial division; GI, gigantocellular reticular nucleus; DI, dysgranular insular cortex; V2MM, secondary visual cortex mediomedial; AuV, secondary auditory cortex, ventral; TeA, temporal association cortex; APTV, anterior pretectal nucleus, ventral.

Table 1.

Key afferent projections to GAD2-positive neurons in the ventrolateral periaqueductal gray: abbreviation glossary.

Abbreviation	Full name
mPFC	Medial prefrontal cortex
cg1	Cingulate cortex 1
cg2	Cingulate cortex 2
LSI	Lateral septal nucleus, intermediate part
Acb	Accumbens nucleus
M1	Primary motor cortex
M2	Secondary motor cortex
S1HL	Primary somatosensory cortex, hindlimb region
LH	Lateral hypothalamic area
CeM	Central amygdaloid nucleus, medial division
GI	Granular insular cortex
DI	Dysgranular insular cortex
V2MM	Secondary visual cx mediom
AuV	Secondary auditory cortex, ventral part
TeA	Temporal association cortex
APTV	Anterior pretectal nucleus, ventral part

In the mPFC region of female and male mice that received simultaneous injections of rAAV2/R-hSyn-FLP-EGFP and rAAV-hSyn-fDIO-mCherry viruses into vlPAG and mPFC, brain slices co-labeled with the excitatory neuron marker CaMKIIα revealed that 83.78% of neurons expressing both EGFP and mCherry also expressed CaMKIIα in male mice and 82.19% in female mice (Fig. 2A). This indicates that most neurons projecting from mPFC to vlPAG are excitatory neurons, with no significant sex differences observed (Fig. 2B, P > 0.05, female vs male). Using the RNAscope technique, we confirmed the specificity of the viral labeling, with 98.66 ± 0.30% and 94.46 ± 2.246% of mCherry-positive neurons co-labeled with Flpo mRNA in female and male mice, respectively (Figs. 2C and D). We further co-labeled RV-DsRed retrogradely traced from vlPAG^GABA to mPFC with CaMKIIα and GABA antibodies in both female and male mice. We observed that neurons retrogradely labeled from vlPAG^GABA in the mPFC predominantly co-expressed CaMKIIα (Fig. 2E) but not GABA (Fig. 2F). Importantly, there were no significant sex differences observed (Fig. 2G, P > 0.05, female vs male).

Figure 2.

Upstream of GABAergic neurons in the vlPAG are predominantly excitatory neurons and show no sex-specific differences. (A) After injecting rAAV2/R-hSyn-FLP-EGFP or rAAV2/9-hSyn-fDIO-Cre-mCherry viruses into the vlPAG and mPFC of WT mice, the mPFC region shows co-labeling of mCherry, EGFP, and CaMKIIα. Scale bar: 100 and 20 μm. Quantification was based on 3 to 5 sections per mouse, with n = 3 male and 3 female mice. (B) Comparison of the proportion of neurons co-labeled with mCherry and EGFP that also express CaMKIIα in female and male mice. Chi-square test, P = 0.6236. (C) Representative images showing the co-expression of RFP (red), GFP (green), and FlpomRNA (blue) in the mPFC of male mice following viral injection. RFP represents mCherry expression, GFP indicates EGFP signal, and FlpomRNA staining confirms the specificity of the viral expression system. Scale bar: 100 and 20 µm. Images were acquired from 3 to 5 sections per mouse, with n = 3 male and 3 female mice. (D) Co-labeling rate of RFP-positive cells in EGFP-positive cells; co-labeling rate of EGFP-positive cells in RFP-positive cells; co-labeling rate of FlpomRNA in RFP-positive cells (n = 3 male and female mice). RFP + GFP/GFP, female, 98.15 ± 0.47%, male, 96.96 ± 0.88%; RFP + GFP/RFP, female, 92.45 ± 0.80%, male, 96.22 ± 0.70%; RFP + FlpomRNA/FlpomRNA, female, 98.66 ± 0.30%, male, 96.46 ± 2.25%. (E) Left, typical image of RV-DsRed co-labeled with CaMKIIα. Right, enlarged view of the boxed region in the left panel. (F) Left, typical image of RV-DsRed co-labeled with GABA. Right, enlarged view of the boxed region in the left panel. (G) Comparison of co-labeling rates of RV-DsRed and CaMKIIα between female and male mice. Chi-square test, P > 0.05. Data were obtained from 3 to 5 sections per mouse, with n = 3 mice per sex, and the average value per animal was used for analysis. n.s, no significant.

3.2. CaMKIIα neurons in the medial prefrontal cortex of female mice express more CB₁R than those in male mice

The mPFC is divided into 5 cell layers,^20,44 with excitatory pyramidal neurons in layers 2/3 (L2/3) and layer 5 (L5) comprising 80% of the total neuron population. Finally, the outgoing projections from L5 and L6 send information from these deep layers to various subcortical regions.⁴⁸ Although previous studies have confirmed the expression of CB₁R on CaMKIIα neurons in the mPFC, it is still unknown whether there are sex differences.^15,30

Given that there is no sex difference in the number of CaMKIIα neurons in the mPFC projecting to vlPAG^GABA in both sexes, we further co-labeled neurons expressing RV-DsRed and CaMKIIα with a CB₁R antibody (Figs. 3A and B). The results indicated that CaMKIIα neurons in the mPFC projecting to vlPAG^GABA express more CB₁R in female mice (Fig. 3C, female vs male, unpaired t test, P < 0.01, n = 4). Given that layer 5 (L5) is the primary output region of the mPFC, we further used fluorescent in situ hybridization to co-localize EGFP with vGlut1 and vGlut2 mRNA in the mPFC of CB₁R-iCre-EGFP mice. Statistical analysis revealed that in layer 5 (L5) of the mPFC, there was no sex difference in the total amount of CB₁R (Figs. 3D–F). However, the total amount of vGlut1+vGlut2 mRNA was higher in male mice than in female mice (Fig. 3F). Despite this, the proportion of vGlut1+vGlut2 mRNA co-labeled with CB₁R was higher in female mice than in male mice (Fig. 3F, female vs male, chi-square test, P < 0.01), further demonstrating the sex-specific distribution of CB₁R on the projections from mPFC^CaMKIIα to vlPAG^GABA.

Figure 3.

The distribution of CB₁R on upstream neurons of GABAergic neurons in the vlPAG shows sex-specific differences. (A) Left, co-labeling of RV-DsRed with CaMKIIα and CB₁R in the mPFC of female mice. Scale bar: 100 μm. Right, magnified view of the boxed region highlighting the co-localization of RV-DsRed and CB₁R. Scale bar: 20 μm. Arrows indicate neurons co-labeled with CB₁R and CaMKIIα. (B) Left, co-labeling of RV-DsRed with CaMKIIα and CB₁R in the mPFC of male mice. Scale bar: 100 μm. Right, magnified view of the boxed region highlighting the co-localization of RV-DsRed and CB₁R. Scale bar: 20 μm. (C) Quantification of the percentage of RV-DsRed-positive neurons that co-express CB₁R within CaMKIIα-positive populations in the mPFC of female and male mice. Data were collected from 12 slices per mouse, n = 4 mice per group. Unpaired t test: female vs male, 0.87 ± 0.03 vs 0.74 ±0.027, t = 3.41, df = 6, *P < 0.05. (D) Co-localization of excitatory neurons (vGlut1+vGlut2 mRNA) and CB₁R-iCre-EGFP in the mPFC region of female and male mice demonstrated by fluorescence in situ hybridization. (E) Enlarged image within the dashed box in panel C. Arrows indicate neurons co-labeled by the vGlut1 + vGlut2 mRNA, CB₁R-EGFP, and DAPI. (F) Comparison of the total number of EGFP (left) and vGlut1 + vGlut2 mRNA (middle) positive neurons between male (n = 6) and female (n = 6) mice. Unpaired t test: female vs male, left, 605.4 ± 47.42 vs 619.9 ± 54.28, t = 0.2012, df =10; middle, 632.3 ± 23.24 vs 831.3 vs 76.49, t = 2.49, df = 10, *P < 0.05; right, comparison of co-labeling rates of vGlut1 + vGlut2 mRNA and CB₁R-EGFP between female and male mice. Data were collected from 3 brain sections per mouse, n = 6 per group; chi-square test, **P < 0.01. ns, no significant.

3.3. Knocking out CB₁R on CaMKIIα neurons in the medial prefrontal cortex exacerbates mechanical allodynia in a sex-specific manner

The aforementioned studies have confirmed the sex-specific distribution of CB₁R on CaMKIIα neurons in the mPFC, but the functional significance of this distribution remains to be explored. We first used RNAscope to examine changes in CB₁RmRNA expression following nerve injury. The results revealed a significant decrease in total CB₁RmRNA levels (Figs. 4A and B, male-WT vs male-CCI, P < 0.001, female-WT vs female-CCI, P < 0.001, 2-way ANOVA). Although a downward trend in CB₁RmRNA expression on CaMKIIα neurons was observed in both sexes, this reduction appeared more pronounced in female mice (Figs. 4C–E, male-WT vs male-CCI, P = 0.057, female-WT vs female-CCI, P < 0.001, sex × treatment interaction, F = 4.815, P = 0.0.043). By specifically knocking down CB₁R in CaMKIIα neurons through injecting rAAV-CaMKIIα-Cre-mCherry virus into the mPFC of CB₁R-flox mice (Figs. 5A–C), we first verified the specificity of CaMKIIα promoter (Figs. 5D and E) and demonstrated that CB₁R was largely knocked out on CaMKIIα neurons in the mPFC (Figs. 5F–G, female, CaMKIIα-Cre-mCherry vs CaMKIIα-mCherry, P < 0.0001, n = 4; male, CaMKIIα-Cre-mCherry vs CaMKIIα-mCherry, P < 0.0001, n = 4), without significant differences in motor function (Fig. 5H, female, CaMKIIα-Cre-mCherry vs CaMKIIα-mCherry, P > 0.05, n = 10; male, CaMKIIα-Cre-mCherry vs CaMKIIα-mCherry, P > 0.05, n = 10). Then, CCI modeling and pain behavioral assessments were conducted on the mice (Fig. 5I). In the physiological condition, there were no significant differences in PWMT, dynamic score, and PTWL (Fig. 5J, Table 2). After nerve injury, both female and male mice with CB₁R knocked out exhibited more pronounced mechanical allodynia compared with the control group (Fig. 5J). Furthermore, Benjamini-Hochberg FDR-corrected 2-way ANOVA revealed localized sex × treatment interactions at 10 days (F = 8.015, @P < 0.05) and 14 d (F = 7.087, @P < 0.05) despite nonsignificant global 3-way interactions (sex × treatment × time: F = 0.55, P = 0.679), indicating sexually dimorphic paw withdrawal mechanical thresholds (PWMTs) during the intermediate recovery phase. Regarding dynamic allodynia scores, while CaMKIIα-Cre-mCherry mice showed significantly enhanced responses compared with controls within each sex, no significant interactions were detected for sex × treatment (2-way ANOVA, P > 0.05), sex × treatment × time (3-way ANOVA, P > 0.05), and timepoint-specific sex × group interactions (Benjamini-Hochberg FDR-corrected, all P > 0.05, Table 2). Similarly, the Hargreaves test revealed no significant effects at any level of analysis (all P > 0.05). Therefore, these results primarily indicate that knocking out CB₁R on all CaMKIIα neurons can exacerbate mechanical allodynia, exhibiting a significant sex difference.

Figure 4.

Reduction of CB₁RmRNA in CaMKIIα neurons after nerve injury in both sexes, with a more pronounced trend in female mice. (A) Representative images showing CB₁RmRNA expression in the mPFC of female and male mice under sham and CCI conditions. (B) Quantification of CB₁RmRNA levels in the mPFC using H-score analysis across different groups, male-WT vs male-CCI vs female-WT vs female-CCI, 155.2 ± 10.01 vs 89.71 ± 4.9 vs 195.1 ± 5.63 vs 118.7 ± 4.185. Data were analyzed by 2-way ANOVA (sex × treatment: F = 0.693, P = 0.417) with sex-stratified tests (female, F = 64.759, P < 0.001; male, F = 49.507, P < 0.001). (C–D) Representative images of CaMKIIα (red), CB₁RmRNA (green), and DAPI (blue) in the mPFC of female (C) and male (D) mice under sham and CCI conditions; Left, low-magnification view of the mPFC; Right, magnified views of dashed-box regions highlighting CaMKIIα-enriched areas. Dashed circles indicate CaMKIIα-enriched areas. Scale bars: 100 μm (left), 20 μm (right). (E) Quantification of CaMKIIα and CB₁RmRNA co-localization using H-score analysis in male and female mice across sham and CCI groups. Male-WT vs male-CCI vs female-WT vs female-CCI, 105.5 ± 4.28 vs 86.32 ± 4.53 vs 164.9 ± 11.31 vs 116.8 ± 2.82. Data were analyzed by 2-way ANOVA (sex × treatment: F = 4.815, P = 0.043) with sex-stratified tests (female, F = 26.591, P < 0.001; male, F = 4.217, P = 0.057). Data are presented as mean ± SEM (n = 5 mice per group, 3-5 brain sections per mouse). Statistical significance was determined using 2-way ANOVA examining sex × treatment interaction, followed by Bonferroni post hoc sex-stratified analysis, *P < 0.05, ***P < 0.001, ****P < 0.0001.

Figure 5.

Knockout of CB₁R on CaMKIIα neurons in the mPFC exacerbates pain in a sex-dependent manner. (A) Diagram of injection of rAAV-CaMKIIα-Cre-mCherry or rAAV-CaMKIIα-mCherry viruses into the mPFC of CB₁R-flox mice. (B) Diagram illustrating the role of the Cre-loxp system in AAV-CaMKIIα-Cre-mCherry virus-mediated knockout of CB₁R. (C) Typical image of rAAV-CaMKIIα-mCherry virus injection into the mPFC. Scale bar: 100 μm. (D) Co-labeling of mCherry (red), CaMKIIα (green), and DAPI (blue) in the mPFC of experimental and control group mice. Scale bar: 100 μm. (E) Descriptive statistics of co-localization between mCherry and CaMKIIα antibody: CaMKIIα-mCherry vs CaMKIIα-Cre-mCherry, 92.66% vs 94.53%. (F) Co-labeling of CaMKIIα (red), CB₁RmRNA (green), and DAPI (blue) in the mPFC of experimental and control group mice. (G) Statistical analysis of H-scores for CB₁RmRNA on CaMKIIα neurons in CaMKIIα-mCherry and CaMKIIα groups across female and male mice. n = 4 mice per group, 3 to 5 brain sections per mouse; unpaired t test, female, CaMKIIα-mCherry vs CaMKIIα-Cre-mCherry, 264.3 ± 11.14 vs 49.29 ± 1.59, t = 19.11, df = 6; male, CaMKIIα-mCherry vs CaMKIIα-Cre-mCherry, 225.5 ± 7.07 vs 47.75 ± 4.40, t = 21.34, df = 6. (H) Female and male mice with CB₁R knockout on CaMKIIα excitatory neurons in the mPFC, and their drop-out latency in the rotarod experiment. unpaired t test, female, 207.8 ± 13.69 vs 222.8 ± 11.45, P = 0.41, t = 0.8406, df = 18; male, 224.7 ± 17.19 vs 205.1 ± 14.82, P = 0.40, t = 0.86, df = 18, CaMKIIα-Cre-mCherry vs CaMKIIα-mCherry. (I) Experimental flowchart. (J) Behavioral changes over time in PWMT, dynamic score, and PWTL after CCI modeling in experimental and control groups of mice; n = 10 to 12, 3-way ANOVA with sex × treatment interactions (von Frey, F = 2.174, P = 0.148; Brush, F = 1.196, P = 0.283; Hargreaves, sex × treatment: F = 0.416, P = 0.522), sex × treatment × time interactions (von Frey, F = 0.55, P = 0.679; Brush, F = 0.402, P = 0.822; Hargreaves, F = 0.585, P = 0.694), and pairwise group comparisons per sex-time combination (*P < 0.05, **P < 0.01, ***P < 0.001, CaMKIIα-Cre-mCherry vs CaMKIIα-mCherry; #P < 0.05, ##P < 0.01, ###P < 0.001, female-CaMKIIα-Cre-mCherry vs female-CaMKIIα-mCherry, &P < 0.05, &&P < 0.01, &&&P < 0.001, male-CaMKIIα-Cre-mCherry vs male-CaMKIIα-mCherry). Two-way ANOVA stratified by time (von Frey: 10 days, F = 8.015, P = 0.044; 14 days, F = 7.087, P = 0.044, @P < 0.05).

Table 2.

Summary of the statistical analysis in Figures 5 and 6.

Figure	Behavior	Statistical test	Comparison groups	P value
Figure 5J	Punctate	Three-way ANOVA	Female-CaMKIIα-Cre-mCherry vs female-CaMKIIα-mCherry	3 d, #P < 0.05; 5 d, ##P < 0.01; 7 d, ###P < 0.001; 10 d, ###P < 0.001; 14 d, ###P < 0.001; 21 d, ###P < 0.001
			Male-CaMKIIα-Cre-mCherry vs male-CaMKIIα-mCherry	7 d, &P < 0.05; 10 d, &&&P < 0.05; 14 d, &&P < 0.01; 21 d, &&P < 0.01
			Female vs male	10 d, @P < 0.05; 14 d, @P < 0.05
	Dynamic	Three-way ANOVA	Female-CaMKIIα-Cre-mCherry vs female-CaMKIIα-mCherry	3 d, ##P < 0.01; 5 d, ##P < 0.01; 7 d, ###P < 0.001; 10 d, ###P < 0.001; 14 d, ###P < 0.01; 21 d, #P < 0.05
			Male-CaMKIIα-Cre-mCherry vs male-CaMKIIα-mCherry	5 d, &P < 0.05; 7 d, &&P < 0.01; 10 d, &&&P < 0.001; 14 d, &&P < 0.01; 21 d, &P < 0.05
			Female vs male	Ns
Figure 6D	Punctate	Three-way ANOVA	Female-CB1R-flox vs female-WT	5 d, ###P < 0.001; 7 d, ###P < 0.001; 10 d, ###P < 0.001; 14 d, ###P < 0.001; 21 d, ###P < 0.001
			Male-CB1R-flox vs male-WT	5 d, &P < 0.05; 7 d, &P < 0.05; 10 d, &&P < 0.01; 14 d, &P < 0.05; 21 d, &&P < 0.01
			Female vs male	7 d, @P < 0.05
	Dynamic	Three-way ANOVA	Female-CB1R-flox vs female-WT	5 d, ###P < 0.001; 7 d, ##P < 0.01; 10 d, ##P < 0.01; 14 d, ###P < 0.001; 21 d, ###P < 0.001
			Male-CB1R-flox vs male-WT	10 d, &&P < 0.01; 14 d, &&P < 0.01; 21 d, &P < 0.05
			Female vs male	Ns
Figure 6G	Punctate	Three-way ANOVA	Female-rAAV2/R-hSyn-CaMKIIα-FLP-CB1R-flox-CCI vs female-rAAV2/R-hSyn-CaMKIIα-FLP-WT-CCI	3 d, ###P < 0.001; 5 d, ###P < 0.001; 7 d, ###P < 0.001; 14 d, ###P < 0.001; 21 d, ###P < 0.001;
			Male-rAAV2/R-hSyn-CaMKIIα-FLP-CB1R-flox-CCI vs male-rAAV2/R-hSyn-CaMKIIα-FLP-WT-CCI	5 d, &P < 0.05; 21 d, &P < 0.05
			Female vs male	3 d, @P < 0.05; 5d, @@P < 0.01; 7d, @@P < 0.01; 10d, @@P < 0.01;14d, @@P < 0.01; 21d, @P < 0.05.
	Dynamic	Three-way ANOVA	Female-rAAV2/R-hSyn-CaMKIIα-FLP-CB1R-flox-CCI vs female-rAAV2/R-hSyn-CaMKIIα-FLP-WT-CCI	3 d, ##P < 0.01; 5 d, ###P < 0.001; 7 d, ###P < 0.001; 10 d, ###P < 0.001; 14 d, ###P < 0.001; 21 d, ###P < 0.001
			Male-rAAV2/R-hSyn-CaMKIIα-FLP-CB1R-flox-CCI vs male-rAAV2/R-hSyn-CaMKIIα-FLP-WT-CCI	5 d, &P < 0.05; 7 d, &&P < 0.01; 10 d, &&&P < 0.001; 14 d, &P < 0.05; 21 d, &P < 0.05
			Female vs male	Ns

^#/&/*P < 0.05,

^##/&&/**P < 0.01,

^###/&&&/***P < 0.001,

^{####/&&&&/****}P < 0.0001.

3.4. Knocking out CB₁R on mPFC^CaMKIIα neurons innervating the ventrolateral periaqueductal gray region exacerbates mechanical allodynia in a sex-specific manner

Given that most neurons projecting from the mPFC to the vlPAG are excitatory, we continued to use the Cre-loxp and Flpo-FRT double system to knock out CB₁R on this long projection and observe the effects on pain behavior. We injected rAAV2/R-hSyn-FLP-EGFP and rAAV2/9-hSyn-fDIO-Cre-mCherry viruses simultaneously into the vlPAG and mPFC regions of CB₁R-flox mice to specifically knock out CB₁R on neurons in the mPFC that project to the vlPAG (Figs. 6A and B). First, we verified that these 2 viruses functioned normally. In WT mice, CB₁R was expressed on neurons co-labeling with EGFP and mCherry in the mPFC projecting to the vlPAG, while in CB₁R-flox mice, CB₁R was largely knocked out (Fig. 6C). Three weeks after virus injection, we measured the baseline behavior of the mice and found no statistical difference between male and female mice (Fig. 6D, Table 2). Following CCI modeling, both male and female mice with CB₁R knocked out exhibited more pronounced mechanical allodynia compared with the control group (Fig. 6D, Table 2). There was a sex difference in PWMT between male and female mice on day 7 (F = 4.546, P = 0.038, @P < 0.05). Regarding dynamic allodynia scores, while CB₁R-flox mice showed significantly enhanced responses compared with controls within each sex, the statistical interactions showed differential significance. Although the sex × treatment interaction was nonsignificant (F = 1.476, P = 0.231, Table 2), we observed a significant 3-way sex × treatment × time interaction (F = 2.557, P = 0.027, Table 2). Notably, thermal nociception assessed by the Hargreaves test showed no significant effects across all analyses (all P > 0.05, Fig. 6D). Furthermore, to further investigate whether similar effects occur when CB₁R is knocked out from CaMKIIα neurons projecting from the mPFC to the vlPAG, we injected rAAV2/R-hSyn-CaMKIIα-FLP into the vlPAG and rAAV-hSyn-fDIO-Cre-mCherry viruses into the mPFC. Using RNAscope, we first confirmed the efficiency of the knockout (Figs. 6E and F) and subsequently observed that female mice exhibited more pronounced punctate pain behavior demonstrated by significant sex × treatment interaction (F = 6.232, P = 0.016), sex × treatment × time interaction (F = 2.679, P = 0.031) and timepoint-specific sex × treatment interactions with Benjamini-Hochberg FDR correction (3 days, F = 4.791, P = 0.034; 5 days, F = 11.772, P = 0.001; 7 days, F = 9.052, P = 0.004; 10 days, F = 9.373, P = 0.004; 14 days. F = 4.479, P = 0.04, @P < 0.05, @@P < 0.01, Fig. 6G). In summary, knocking out CB₁R on CaMKIIα neurons in the mPFC projecting to the vlPAG induces more pronounced mechanical allodynia in female mice.

Figure 6.

Selectively knocking out CB₁R on CaMKIIα excitatory neurons in the mPFC region using the Cre-loxp and Flpo-FRT dual recombinase system enhanced mechanical pain. (A) Schematic diagram of rAAV2/R-hSyn-FLP-EGFP or rAAV2/9-hSyn-fDIO-Cre-mCherry virus injection into vlPAG and mPFC of WT or CB₁R-flox mice. (B) Experimental flowchart of virus injection, CCI modeling, and behavioral testing. (C) mPFC region of WT and CB₁R-flox mice showing mCherry, EGFP, and CB₁R co-labeling. Scale bar: 50 μm. Arrows indicate projection neurons lacking CB₁R. (D) Behavioral changes over time in PWMT, dynamic score, and PWTL after CCI modeling in WT and CB₁R-flox mice. n = 6 to 14, 3-way ANOVA with sex × treatment interactions (von Frey, F = 2.36, P = 0.131; Brush, F = 1.476, P = 0.231; Hargreaves, sex × treatment: F = 1.58, P = 0.215), sex × treatment × time interactions (von Frey, F = 1.37, P = 0.248; Brush, F = 2.557, P = 0.027; Hargreaves, F = 1.491, P = 0.206), and pairwise group comparisons per sex-time combination (*P < 0.05, **P < 0.01, ***P < 0.001, WT-CCI vs CB₁R-flox-CCI. #P < 0.05, ##P < 0.01, ###P < 0.001, female-WT-CCI vs female-CB₁R-CCI, &P < 0.05, &&P < 0.01, &&&P < 0.001, male-WT-CCI vs male-CB₁R-CCI). Two-way ANOVA stratified by time (von Frey, 7 d, F = 4.546, P = 0.038, @P < 0.05). (E) Representative images of the mPFC region in WT and CB₁R-flox mice showing co-labeling of RFP (red), CB₁RmRNA (green), and DAPI (blue). Viral injection sites: vlPAG, rAAV2/R-hSyn-CaMKIIα-FLP, mPFC, rAAV-hSyn-fDIO-Cre-mCherry. (F) H score of CB₁RmRNA of the RFP positive neurons in the mPFC of WT mice and CB₁R-flox mice. n = 4 mice per group, 3 to 5 brain sections per mouse; unpaired t test. WT vs CB₁R-flox, 227.6 ± 6.84 vs 92.20 ± 7.93, ****P < 0.0001, t = 12.93, df = 6. (G) Behavioral changes over time in PWMT, dynamic score, and PWTL after CCI modeling in WT and CB₁R-flox mice receiving rAAV2/R-hSyn-CaMKIIα-FLP virus in the vlPAG and rAAV-hSyn-fDIO-Cre-mCherry in the mPFC. n = 6 to 12, 3-way ANOVA with sex × treatment interactions (von Frey, F = 6.232, P = 0.016; Brush, F = 4.937, P = 0.031; Hargreaves, sex × treatment: F = 2.822, P = 0.1), sex × treatment × time interactions (von Frey, F = 2.679, P = 0.031; Brush, F = 1.881, P = 0.097; Hargreaves, F = 0.223, P = 0.929), and pairwise group comparisons per sex-time combination, *P < 0.05, **P < 0.01, ***P < 0.001, WT-CCI vs CB₁R-flox-CCI. #P < 0.05, ##P < 0.01, ###P < 0.001, female-WT-CCI vs female-CB₁R-CCI, &P < 0.05, &P < 0.01, &P < 0.001, male-WT-CCI vs male-CB₁R-CCI). Two-way ANOVA stratified by time (von Frey: 3 days, F = 4.791, P = 0.034; 5 days, F = 11.772, P = 0.001; 7 days, F = 9.052, P = 0.004; 10 days, F = 9.373, P = 0.004; 14 days, F = 4.479, P = 0.04, @P < 0.05, @@P < 0.01).

3.5. Specific activation of CB₁R on mPFC^CaMKIIα neurons can yield stronger analgesic effects in female mice

As specific knockout of CB₁R on CaMKIIα neurons in the mPFC induces more severe pain in female mice, we hypothesize that specific activation of this receptor may produce stronger analgesic effects in female mice. We first examined the effects of systemic administration of the CB₁R agonist ACEA on neuropathic pain. The results showed that, on day 7 post-CCI (Fig. 7A), intraperitoneal injection of ACEA significantly alleviated mechanical allodynia with more pronounced effects in female mice (Fig. 7B), supported by 3-way ANOVA showing significant sex × treatment interaction (von Frey, F = 36.855, P < 0.001; Brush, F = 6.549, P = 0.019) and sex × treatment × time interaction (von Frey, F = 8.876, P < 0.001; Brush, F = 3.802, P = 0.003, n = 6 mice per group). Then, we began to explore the functions of specifically activating CB₁R on CaMKIIα neurons in the mPFC. To specifically activate CB₁R in CaMKIIα excitatory neurons, we utilized the opto-XR method by employing a photosensitive chimeric receptor based on CB₁R. ³⁸ This receptor was engineered by substituting the intracellular loops and the C terminus of rhodopsin with those of human CB₁R (Fig. 7C). We introduced this opto-CB₁R into the CaMKIIα interneurons in the mPFC of CCI mice using a viral vector and implanted an optic fiber to deliver blue laser stimulation (Figs. 7D and E). Based on previous reports,^13,38 we chose a 20-Hz pulse stimulation for the optogenetic activation of opto-CB₁R. On the seventh day after nerve injury, when the mice exhibited obvious mechanical allodynia, we first measured their baseline. Both female and male mice then received 2 minutes of blue light stimulation (20 Hz, 20 ms). Pain behavior was assessed at 2 minutes, 20 minutes, 30 minutes, 1 hour, 2 hours, and 3 hours poststimulation (Figs. 6F and G). The results showed significant relief of both punctate and dynamic mechanical allodynia at 2 minutes, 20 minutes, and 30 minutes, with 3-way ANOVA confirming robust sex-dependent effects: significant sex × treatment interactions (von Frey: F = 16.232, P < 0.001; Brush, F = 12.682, P < 0.001) and sex × treatment × time interactions (von Frey, F = 5.202, P = 0.001; Brush, F = 5.013, P < 0.001 n = 8 mice per group), indicating that specific activation of CB₁R on CaMKIIα neurons in the mPFC produced more pronounce analgesic effects in female mice.

Figure 7.

Specific activation of CB₁R in the mPFC alleviates pain in a sex-specific manner. (A) Flowchart of CCI nerve injury, ACEA administration, and behavioral assessment. (B) Behavioral changes in von Frey and Brush test at 7 days post-CCI and at 1 hour, 2 hours, 4 hours, 6 hours, 8 hours, 1 d, 3 d, and 5 d after intraperitoneal injection of 10 mg/kg ACEA. n = 6, 3-way ANOVA with sex × treatment interactions (von Frey, F = 36.855, P < 0.001; Brush, F = 6.549, P = 0.019), sex × treatment × time interactions (von Frey, F = 8.876, P < 0.001; Brush, F = 3.802, P = 0.003), and pairwise group comparisons per sex-time combination (*P < 0.05, **P < 0.01, ***P < 0.001, Vehicle vs ACEA; #P < 0.05, ##P < 0.01, ###P < 0.001, female-Vehicle vs female-ACEA; &P < 0.05, &&P < 0.01, &&&P < 0.001, male-Vehicle vs male-ACEA). Two-way ANOVA stratified by time (von Frey: 1 hour, F = 29.702, P < 0.001; 2 hours, F = 27.616, P < 0.001; 4 hours, F = 14.997, P = 0.00237; 6 hours, F = 28.864, P = 0.000127; Brush: 2 hours, F = 7.778, P = 0.0275; 4 hours, F = 11.942, P = 0.02; Brush: @@P < 0.01, @@@P < 0.001). (C) Diagram of optogenetic manipulation of opto-CB₁R expression. (D) Schematic diagram of pAAV-CaMKIIα-opCB₁R-mCherry virus injection into the mPFC of WT mice and blue light stimulation through optical fibers in vlPAG. (E) Diagram illustrating the expression of CaMKIIα-opCB₁R-mCherry in the mPFC and its projections in vlPAG. (F) Flowchart of virus injection, CCI modeling, optical fiber implantation, and behavioral assessment with von Frey and Brush tests at 2 minutes, 20 minutes, 30 minutes, 1 hour, 2 hours, and 4 hours after blue light stimulation (473-nm laser, 20 Hz with 20 ms, 2 minutes). (G) Behavioral changes in the von Frey and Brush tests at 7 days post-CCI and at 2 minutes, 20 minutes, 30 minutes, 1 hour, 2 hours, and 3 hours after blue light stimulation. n = 8, 3-way ANOVA with sex × treatment interactions (von Frey, F = 16.232, P < 0.001; Brush, F = 12.682, P < 0.001), sex × treatment × time interactions (von Frey, F = 5.202, P = 0.001; Brush, F = 5.013, P < 0.001), and pairwise group comparisons per sex-time combination (*P < 0.05, **P < 0.01, ***P < 0.001, CCI + AAV-CaMKIIα-mCherry vs CCI + AAV-CaMKIIα-opCB₁R-mCherry. #P < 0.05, ##P < 0.01, ###P < 0.001, female-CCI + AAV-CaMKIIα-mCherry vs female-CCI + AAV-CaMKIIα-opCB₁R-mCherry; &P < 0.05, &&P < 0.01, &&&P < 0.001, male-CCI + AAV-CaMKIIα-mCherry vs male-CCI + AAV-CaMKIIα-opCB₁R-mCherry). Two-way ANOVA stratified by time (von Frey: 2 minutes, F = 15.286, P = 0.00214; 20 minutes, F = 17.062, P = 0.00214; 30 minutes, F = 12.58, P = 0.00372; Brush: 2 minutes, F = 15.154, P = 0.00238; 20 minutes, F = 14.976, P = 0.00238; 30 minutes, F = 12.077, P = 0.004483).

3.6. Activation of CB₁R on mPFC^CaMKIIα neurons more significantly reduces the amplitude of oEPSCs recorded from vlPAG^GABA of female mice

Next, we investigated the mechanisms underlying the enhanced analgesic effects of selectively activating CB₁R on CaMKIIα neurons in the mPFC of female mice. Activation of CB₁R on GABA neurons is the main mechanism through which cannabinoids exert their analgesic effects in the vlPAG.²¹ We first investigated how the sex-specific distribution of CB₁R on mPFC^CaMKIIα influences synaptic transmission in GABAergic neurons within the vlPAG. Then, we initially injected the rAAV-CaMKIIα-ChR2-EGFP virus into the mPFC of GAD2-Cre-tdTomato mice (Fig. 8A). After 3 weeks, we prepared brain slices and performed whole-cell patch-clamp recordings of CaMKIIα-EGFP neurons in the mPFC. These recordings showed action potentials in response to 20-Hz blue light stimulation, confirming functional expression of ChR2 (Figs. 8B and C).

Figure 8.

Sex-specific reduction of oEPSC in the mPFC^CaMKIIα-vlPAG^GABA circuit by selectively activating CB₁R on mPFC^CaMKIIα. (A) Schematic diagram of rAAV2/9-CaMKIIα-ChR2-EGFP virus injection into the mPFC region of female or male GAD2-tdTomato mice. (B) Optogenetic and electrophysiological experimental setup diagram. (C) Blue light stimulation of mPFC^CaMKIIα neurons induces action potentials in mPFC. (D) Axon terminals of CaMKIIα neurons distributed around GAD2-tdTomato neurons within the vlPAG, scale bar, 100 μm. (E) Representative image of oEPSC before and after 10 μM ACEA perfusion in female mice. (F) Representative image of oEPSC before and after 10 μM ACEA perfusion in male mice. (G) The change in oEPSC amplitude induced by blue light stimulation in female mice following ACEA perfusion. n = 12 to 13 neurons from 5 mice, 2-way ANOVA with sex × treatment interactions (F = 4.651, P = 0.0363) followed by post hoc (****P < 0.0001, baseline vs ACEA).

Simultaneously, we verified viral expression. The results revealed EGFP-positive CaMKIIα neurons in the mPFC and ChR2-expressing projection fibers from the mPFC to the vlPAG, surrounding GABAergic neurons in the vlPAG, indicating successful viral expression (Fig. 8D). In the vlPAG, we recorded both opto-inhibitory optogenetically evoked inhibitory postsynaptic current (oIPSCs) and optogenetically evoked excitatory postsynaptic current (oEPSCs). The data showed that GABAergic neurons in the vlPAG predominantly receive excitatory input from CaMKIIα mPFC neurons, as indicated by the oEPSCs. Perfusion with 10 μM ACEA (a CB₁R agonist) significantly reduced the amplitude of oEPSCs in both female and male mice, with a notable sex difference (Figs. 8E–G, 2-way ANOVA with sex × treatment interactions, F = 4.651, P = 0.0363, n = 12-13 neurons from 5 mice). These results indicate that stimulation of CB₁R on CaMKIIα neurons in the mPFC leads to a more pronounced decrease in the amplitude of oEPSCs recorded from GABAergic neurons in the vlPAG of female mice.

Activation of CB₁R on mPFC^CaMKIIα neurons more significantly reduces the function of the GABA-CaMKIIα feedforward inhibition circuit in the vlPAG of female mice.

In the vlPAG, based on the primary lateral inhibition model, local GABAergic neurons mainly project to the glutamatergic output neurons, thereby modulating the PAG-RVM descending inhibitory pathway. Cannabinoids exert analgesic effects by indirectly exciting output neurons through the inhibition of GABAergic neurons.²¹ The glutamatergic output neurons in the vlPAG play a crucial role in descending pain modulation. Therefore, after investigating the impact of CB₁R activation on mPFC^CaMKIIα-vlPAG^GABA projection, we further examined its effect on synaptic transmission activities of CaMKIIα output neurons. We injected pAAV-CaMKIIα-EGFP and pAAV-CaMKIIα-opCB₁R-mCherry viruses into the vlPAG and mPFC of wild-type male and female mice (Fig. 9A). Fourteen days postinjection, we performed CCI modeling and prepared brain slices on the 21st day.

Figure 9.

Sex-specific reduction of GABA-vGlut2 sIPSC by selectively activating CB₁R on CaMKIα, with no effect on excitatory transmission from mPFC^CaMKIIα to vGlut2^vlPAG. (A) Schematic diagram of pAAV-CaMKIIα-opCB₁R-mCherry or pAAV-CaMKIIα-EGFP virus injection into mPFC and vlPAG of WT mice in both sexes. (B) Optogenetic and electrophysiological experimental setup diagram. (C) Typical diagram of recording CaMKIIα neurons in the vlPAG. (D) Representative sIPSC traces before (OFF) and after (ON) blue light stimulation in female mice. (E) Representative sIPSC traces before (OFF) and after (ON) blue light stimulation in male mice. (F) The frequency change of sIPSCs before and after blue light stimulation in the female and male mice. Blue light stimulation of CB₁R on CaMKIIα neurons of mPFC induced a more robust reduction of frequency of sIPSCs in the female mice than in male mice. n = 13 neurons from 5 mice per group, 2-way ANOVA with sex × treatment interactions (F = 4.486, P = 0.0394) followed by post hoc (****P < 0.0001, OFF vs ON). (G) There was no difference in the amplitude of sIPSCs under OFF and ON conditions in female and male mice. n = 13 neurons from 5 mice per group, 2-way ANOVA with sex × treatment interactions (F = 0.04058, P = 0.8412) followed by post hoc (n.s, no significant, OFF vs ON). (H) Representative sEPSCs traces under OFF and ON conditions in female mice. (I) Representative sEPSCs traces under OFF and ON conditions in male mice. (J) There was no difference in the frequency of sEPSCs under OFF and ON conditions in female and male mice. n = 12 neurons from 5 mice per group, 2-way ANOVA with sex × treatment interactions (F = 1.002, P = 0.3224) followed by post hoc (n.s, no significant, OFF vs ON). (K) There was no difference in the amplitude of sEPSCs under OFF and ON conditions in female and male mice. n = 12 neurons from 5 mice per group, 2-way ANOVA with sex × treatment interactions (F = 0.5258, P = 0.4722) followed by post hoc (n.s, no significant, OFF vs ON).

Using electrophysiology combined with optogenetics, we clamped EGFP neurons in the vlPAG (Figs. 9B and C). First, at a holding potential of 0 mV, we recorded sIPSCs before and after blue light stimulation in both male and female mice. The results showed that the frequency of sIPSCs decreased after light stimulation mainly in female mice (Figs. 9D-F, female, OFF vs ON, P < 0.0001, n = 13 neurons from 5 mice; male, OFF vs ON, P =0.117, n = 13 neurons from 5 mice), with a more pronounced reduction in female mice (Figs. 9D-F, 2-way ANOVA with sex × treatment interactions, F = 4.651, P = 0.0363, n = 13 neurons from 5 mice). The amplitude of sIPSCs remained unchanged in both sexes after light stimulation (Figs. 9D, E, and G). In addition, the sEPSCs were recorded at a holding potential of −70 mV, with results showing that the frequency and amplitude of sEPSCs did not significantly change after blue light stimulation in either sex (Figs. 9H-K, P > 0.05, n = 12 neurons from 5 mice). These findings indicate that the activation of CB₁R on mPFC^CaMKIIα neurons induces synaptic plasticity changes in CaMKIIα output neurons in the vlPAG primarily through influencing the function of feedforward inhibition circuits, rather than through direct excitatory action.

4. Discussion

This study showed that the retrograde projection of vlPAG^GABA into the mPFC is predominantly CaMKIIα⁺ excitatory neurons, and abundant CB₁Rs were distributed in the mPFC^CaMKIIα-vlPAG^GABA circuit. Activation of CB₁R in mPFC^CaMKIIα-vlPAG^GABA circuit reduces excitatory synaptic responses, and GABA neurons within vlPAG are relatively inhibited. Finally, vlPAG^GABA-vlPAG^vGlut2 inhibitory and mPFC^CaMKIIα-vlPAG^GABA excitatory circuit synergistically promoted the analgesic effect of cannabinoids (Fig. 10). The above findings revealed sex differences in cannabinoid-mediated analgesia, which provided a structural and functional basis for exploring more specific therapeutic approaches for chronic pain in female mice.

Figure 10.

Working hypothesis of the mPFC^CaMKIIα-vlPAG^GABA circuit. CB₁R expression on mPFC^CaMKIIα neurons specifically projecting to vlPAG^GABA is higher in female mice (A) than in male mice (B). Compared with male mice, specific activation of CB₁R on mPFC^CaMKIIα terminals more significantly reduces the oEPSC recorded in vlPAG^GABA neurons and the sIPSCs recorded in vlPAG^vGlut2 neurons in female mice, while also more significantly enhancing the synaptic plasticity of vGlut2 neurons in female mice. This leads to sex differences in the endogenous analgesic effects of the endocannabinoid system, producing a stronger therapeutic effect on neuropathic pain in female mice. Glu, glutamate; vGlut2, vesicular glutamate transporter 2. Pink: mPFC region; blue: vlPAG region; purple: CaMKIIα neuron; orange: GABAergic neuron; green: vGlut2 neuron.

Neurons in the mPFC region exhibit a high expression of CB₁R, and this region also shows a sex-specific distribution of CB₁R.³ Liu et al. reported that the CB₁R-mRNA in the anterior cingulate cortex (ACC) of male mice was higher than in female mice, as determined by in situ hybridization.²⁶ Conversely, Xing et al. found that CB₁R-mRNA levels in the PFC of female mice were much higher than in male mice.⁵⁰ However, neither study investigated the cell-specific distribution of CB₁R in this region. Previous research generally suggested that CB₁R in the mPFC is predominantly expressed on GABAergic neurons,⁴⁵ particularly those positive for calbindin but not parvalbumin or calretinin. This view is not entirely accurate, as Elisa L. Hill and Giovanni Marsicano et al. discovered significant CB₁R expression on pyramidal neurons.^15,30 In our study, we also observed substantial CB₁R expression on CaMKIIα excitatory neurons in the mPFC. Moreover, the expression of CB₁R-mRNA on CaMKIIα neurons was higher in female mice compared with male mice. These findings, along with the focus on sex-specific GPCR distribution in pain-related brain regions,²⁴ may enrich our understanding of CB₁R distribution on CaMKIIα neurons in the mPFC, while also aiding in the development of effective analgesics for both sexes and deepening insights into sex-specific cannabinoid analgesia.

Current research on the endocannabinoid system in the mPFC mainly focuses on the effects of activating CB₁R on local GABAergic neurons and their inhibitory influence on postsynaptic excitatory neurons, with particular emphasis on interactions between the mPFC and the amygdala.^16,49 CaMKIIα-positive pyramidal neurons in the mPFC, which serve as projection neurons, also express CB₁R. Despite limited previous research on CB₁R expression in projections across brain regions, our experiments reveal critical insight: following nerve injury, CB₁RmRNA levels in the mPFC are markedly reduced in both female and male mice, with a more pronounced downregulation of CB₁R on CaMKIIα-expressing neurons in female mice. To functionally dissect this observation, we knocked out CB₁R specifically in presynaptic CaMKIIα neurons within the mPFC^CaMKIIα-vlPAG^GABA circuit, which significantly exacerbates mechanical allodynia in both sexes. This finding highlights the analgesic role of CB₁R on CaMKIIα neurons in the mPFC and suggests its involvement in modulating presynaptic neurotransmitter release to gate pain signaling. The stronger CB₁R-dependent phenotype in female mice may arise from estrogen modulation, which enhances CB₁R coupling to Gi/o signaling pathways, amplifying its regulatory impact on synaptic transmission in female mice.¹⁹ Furthermore, it is also worth exploring whether the specific activation of CB₁R on mPFC^CaMKIIα is influenced by the estrous cycle in female mice in the future. It may also be attributed to endocannabinoid metabolic bias, in which sex-specific shifts in 2-AG/AEA synthesis or degradation postinjury could drive greater compensatory CB₁R downregulation in female mice.¹¹ Profiling endocannabinoid levels and CB₁R coupling efficiency in the vlPAG across sexes could unravel metabolic and signaling bases for these disparities. These findings align with Weizman et al.'s report that CB₁R activation reduces connectivity in chronic pain networks of the dlPFC (analogous to the mPFC in rodents) to produce analgesia.⁴⁷ It is worth noting that our use of an opto-CB₁R strategy enabled precise and direct activation of CB₁R on mPFC^CaMKIIα neurons, circumventing off-target effects of systemic CB₁R agonists.³² This approach not only clarifies CB₁R's circuit-specific role but also highlights its therapeutic potential, particularly given the sex-dependent CB₁R dynamic observed in this study, which may necessitate sex-tailored interventions for neuropathic pain.

Although our study primarily addresses nociceptive hypersensitivity, the mPFC-vlPAG circuit's role in affective and cognitive pain processing warrants special emphasis. The vlPAG serves as a nexus not only for pain modulation but also for aversive emotional states through its reciprocal connections with limbic regions and the rostral ventromedial medulla.^16,51 Patients experiencing chronic pain exhibit mPFC-dependent working memory deficits.³⁵ Our observation of sex-specific CB₁R distribution in the mPFC^CaMKIIα-vlPAG^GABA circuit raises an important question: Could the female-biased CB₁R downregulation after nerve injury predispose female mice to stronger comorbidity between mechanical allodynia, affective disturbances, and cognitive dysfunction? It is also known that the mPFC can be divided into the prelimbic cortex (PrL) and infralimbic cortex (IL) subregions, which often have differing functions.⁴¹ Hence, to further elucidate the functional organization of CB₁R-mediated pain modulation, future studies should investigate subregional specificity by examining CB₁R expression and function in CaMKIIα neurons across distinct mPFC subregions; integrate affective-cognitive assessments with existing nociceptive tests through conditioned place aversion (CPA) and T-maze alternation tasks to quantify pain-related aversion and cognitive dimension; employ projection-specific CB₁R knockout combined with miniScope imaging to track vlPAG^GABA neuronal activity during affective behaviors; and test whether CB₁R deletion in this circuit exacerbates pain-related emotional deficits and cognitive dysfunction in a sex-dependent manner.

The absence of recorded effects of CB₁R modification in mPFC^CaMKIIα-vlPAG ^GABA-vlPAG^vGlut2 on thermal hyperalgesia, as evaluated by the Hargreaves test, exposes an interesting dissociation between tactile hypersensitivity and thermal hyperalgesia. The basic variations in the sensory pathways and mechanisms underlying these 2 forms of pain may help to explain this disparity. Mostly from non-noxious stimuli like touch, tactile hypersensitivity results from sensory information being transferred by myelinated fibers (eg, Aβ fibers) from the dorsal columns (DCs) to the spinal dorsal horn and subsequently sent to higher brain regions, including the thalamus. On the other hand, thermal hyperalgesia—induced by noxious stimuli—relies on unmyelinated C fibers and somewhat myelinated Aδ fibers to carry information to projection neurons in the spinal dorsal horn and then to brain areas like the parabrachial nucleus (PB).^1,27 This variation in sensory processing may influence the varied outcomes seen in our work.

Although other studies have shown that mPFC^CaMKIIα neurons can induce heat hyperalgesia,¹⁶ various elements could help to explain the lack of effects in our particular experimental model. First, depending on the damage model utilized, tactile hypersensitivity and thermal hyperalgesia usually show different degrees of manifestation.⁴⁰ The model of chronic constriction injury (CCI) uses ligation instead of nerve transection, which by default adds variability in the sensory phenotype of the model. Subtle variations in the degree of nerve damage among studies could affect the predominance of thermal hyperalgesia or tactile hypersensitivity. Furthermore, the selective influence on tactile hypersensitivity could be explained by the specificity of our intervention aiming at CB₁R on CaMKIIα neurons instead of the whole CaMKIIα neuronal population. The type of sensory input and the engaged downstream brain circuits will affect CB₁R-mediated modulation of pain. Evidence of functional segregation of GPCR-mediated signaling in pain control supports this theory by means of opioid receptors.³⁷ Different effects on mechanical pain and heat pain have been revealed by opioid receptor-mediated analgesia, implying that molecular variations in receptor-mediated signaling could be responsible for the dissociation shown in our work. Future studies are justified to investigate the microcircuit-level processes by which CB₁R expressed on mPFC^CaMKIIα neurons preferentially controls tactile hypersensitivity but not heat hyperalgesia. These investigations should seek to identify the downstream neural circuits activated by mPFC^CaMKIIα neurons in response to various sensory modalities as well as the molecular and cellular processes driving GPCR-mediated pain segregation. Knowing these pathways could help one to better understand the varied control of tactile and thermal pain and guide therapy plans for specific pain management. The subcellular location of CB₁R and its sex-dependent distribution in the mPFC-vlPAG circuit explore the cortical-subcortical circuits of the endogenous pain control system overall. It investigates further how particular activation of CB₁R controls the activity of the GABA-CaMKIIα feedforward inhibitory circuit within the vlPAG, therefore mediating sex-specific effects of cannabis treatment on neuropathic mechanical allodynia. The results set the foundation for creating analgesic treatments suitable for both men and women and offer first hints on the processes behind sex variations in the analgesic effects of GPCR-related medicines, best shown by CB₁R.

Overall, this study investigates the cortical-subcortical circuits of the endogenous pain control system, focusing on the subcellular localization of CB₁R and its sex-dependent distribution in the mPFC-vlPAG circuit. It further examines how specific activation of CB₁R regulates the function of the GABA-CaMKIIα feedforward inhibitory circuit within the vlPAG, thereby mediating sex-specific effects of cannabinoid treatment on neuropathic mechanical allodynia. The findings provide preliminary insights into the mechanisms underlying sex differences in the analgesic effects of GPCR-related drugs, exemplified by CB₁R, and lay the groundwork for developing analgesic treatments effective for both men and women.

Conflict of interest statement

The authors have no conflicts of interest to declare.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/C418.

Supplementary Material

jop-167-691-s001.pdf ^{(501.3KB, pdf)}