The EMR exposure device employed in this investigation is illustrated in Figures 1A,B. Mice in the radiation group, situated within an animal radiation box (Bai li Acrylic Products Factory, Beijing, China), underwent irradiation within the effective work area of the RC. In contrast, the control group mice were also placed in a radiation box but were not subjected to irradiation to eliminate the influence of irrelevant variables.

To explore the frequency-dependent effects on mice behavior, single-frequency models of 2.65 or 0.8 GHz EMR with a specific absorption ratio (SAR) at 4 W/kg were established. The process of establishing these EMR models and conducting behavioral evaluations is depicted in Figure 1C. The exposure duration was 4 h per day for a total of 21 days, excluding days designated for behavioral testing (e.g., days 7, 14, and 21). The results demonstrated that, when compared with the control group, the single-frequency 2.65 or 0.8 GHz EMR groups exhibited no significant differences in center distance and center time in open field test (OFT) on day 7, 14 and 21 (p > 0.05, Figures 1D–G,L–O). Additionally, there were no significant differences in the distance time and entries into the open arm in the elevated plus maze (EPM) on day 7, 14, and 21 (p > 0.05, Figures 1H–K,P–S). These findings collectively indicate that prolonged exposure to single-frequency EMR (2.65 or 0.8 GHz) did not induce anxiety-like behavior in mice.

To investigate the impact of multi-frequency EMR on brain function, we established a dual-frequency (2.65/0.8 GHz) EMR model, exploring both behavioral effects and molecular mechanisms (Figure 2A). Mice were randomly assigned to control and dual-frequency 2.65/0.8 GHz EMR groups. The dual-frequency group underwent exposure to 2.65 GHz radiation for 2 h, followed by 0.8 GHz radiation for an additional 2 h, totaling 4 h per day. The total radiation time (4 h) and SAR (4 W/kg) per day in the dual-frequency radiation group matched those in the single-frequency radiation group.

Mouse weights were measured during the radiation, revealing no differences between the control and dual-frequency (2.65/0.8 GHz) EMR groups (Figure 2B). Additionally, the rectal temperature of mice was monitored using a fiber-optic thermometer during dual-frequency (2.65/0.8 GHz) EMR exposure (Figure 2C), with ambient temperature at 18.6°C and humidity at 21.5%. The mean rectal temperature of mice before radiation was 36.0788°C, reaching a mean highest temperature of 36.3934°C and a mean lowest temperature of 36.0812°C (Figures 2D,E). This demonstrates that exposure to dual-frequency (2.65/0.8 GHz) EMR with SAR of 4 W/kg for 4 h caused a temperature increase of less than 1°C, allowing the research to focus on non-thermal effects while ruling out thermal effects in mice.

In the OFT (Figures 2F–H), the dual-frequency EMR group exhibited a significant reduction in center distance (Supplementary Figure 1A; p = 0.0217, day 7; p = 0.0094, day 14; p = 0.0382, day 21) and center time (Figure 2H; p = 0.0259, day 7; p = 0.0249, day 14; p = 0.0469, day 21) on day 7, 14, and 21 compared to the control group. In the EPM (Figures 2I–K; Supplementary Figure 1B), the dual-frequency EMR group displayed a significant reduction in distance (Figure 2J; p = 0.0402, day 7; p = 0.0293, day 14; p = 0.0111, day 21) and time (Figure 2K; p = 0.0307, day 7; p = 0.0231, day 14; p = 0.0448, day 21) in open arms on the same days. These findings indicate that dual-frequency (2.65/0.8 GHz) EMR induced anxiety-like behavior in mice. However, in comparison with the control group, the dual-frequency EMR group showed no significant difference in the total immobility duration during the tail suspension test (TST) on days 7, 14, and 21 (Figure 2L), suggesting that dual-frequency (2.65/0.8 GHz) EMR did not induce depression-like behavior in mice.

Furthermore, we analyzed serum hormones in mice, revealing no significant difference in norepinephrine (NE) content (p > 0.05, Figure 2M). However, noteworthy increases were observed in corticosterone (CORT) and corticotrophin releasing hormone (CRH) content (p = 0.0349, Figure 2N; p = 0.0078, Figure 2O) in the dual-frequency EMR group compared with the control group. These findings suggest that the anxiety-like behavior induced by dual-frequency EMR (2.65/0.8 GHz) may be related to alterations in CORT and CRH levels within the HPA axis.

In this part of the study, we established that prolonged exposure to dual-frequency 2.65/0.8 GHz EMR, below the thermal threshold (≤ 1°C), exerts a detrimental impact on both emotion and serum hormones in mice. This impact includes the induction of anxiety-like behavior without depression-like behavior, along with an upregulation of CORT and CRH content.

To investigate the molecular mechanisms triggering anxiety-like behaviors due to dual-frequency EMR, we extracted total RNA from the brain cortex and hippocampus of mice. We conducted an RNA-sequencing analysis (Figure 3A). The correlation of gene expression levels among all sequenced samples surpassed 0.92 (R² > 0.92 represents ideal sampling and experimental conditions), indicating ideal sampling conditions and biological repeatability (Figure 3B).

In the brain cortex, 1869 differentially expressed genes (DEGs) were identified, consisting of 860 upregulated and 1,009 downregulated genes (Figure 3C). Kyoto encyclopedia of genes and genomes (KEGG) metabolic pathway analysis highlighted 12 significantly different metabolic pathways with a focus on the retrograde endocannabinoid signaling pathway (p = 6.41 × 10⁻¹⁴, consistent with the endocannabinoid system, ECS), closely linked to emotion regulation in mice (Figure 3D). Within this pathway, 51 DEGs were identified, including 10 upregulated and 41 downregulated DEGs. Compared to the control group, the Cnr1 gene encoding CB1R was notably downregulated in the radiation group (p = 0.025, Figure 3E). Western blotting confirmed a significant reduction in CB1R protein expression in the cortex (p = 0.0367, Figure 3F), indicating that prolonged exposure to dual-frequency EMR impacted the CB1R expression of the ECS within the mice’s brain cortex. Similar analyses in the brain hippocampus revealed a significant difference in the retrograde endocannabinoid signaling pathway but no significant difference in Cnr1 expression (Supplementary Figures S2A–C; p = 0.025, Supplementary Figure S2B). Western blotting for CB1R protein expression in the hippocampus and thalamus showed no difference in the dual-frequency EMR group compared to the control group (Figure 3F).

In summary, these results suggest that the downregulation of CB1R in the ECS within the brain cortex may lead to decreased binding with corresponding ligands, resulting in the downregulation of ECS function, ultimately leading to anxiety-like behavior in mice.

Pertaining to the endocannabinoids 2-AG and AEA, crucial ligands of the ECS that predominantly bind to CB1R and regulate mood, we measured their molecular levels with an ELISA. Our analysis revealed a significant reduction in the levels of the 2-AG molecule in the cortex (p = 0.033, Figure 4A) and no significant change in the levels of the AEA molecule (p > 0.05, Figure 4B) in mice exposed to dual-frequency EMR compared to the control group. Furthermore, we investigated the enzymes responsible for controlling endocannabinoid levels in the brain, specifically MGL, which regulates 2-AG, and FAAH, which regulates AEA. However, our investigations did not unveil any significant alterations in the levels of MGL or FAAH enzymes (p > 0.05, Figures 4C,D).

These findings strongly suggest that exposure to dual-frequency EMR leads to the downregulation of the 2-AG ligand within the ECS of the brain cortex while leaving the levels of MGL and FAAH enzymes unaffected.

Given the pivotal role of the ECS in emotion regulation and the observed changes in ECS activity in the brains of mice exposed to dual-frequency EMR in this study, our next objective was to investigate whether increasing ECS activity could mitigate neurobehavioral harm. Starting from the 8th to 21st day post-EMR exposure, mice in the RW group received daily intraperitoneal injections of Win55-212-2, a cannabinoid receptor agonist obtained from Shanghai Yi Xi Chemical Technology Co., LTD, at a dose of 1 mg/kg. To provide context, this meant that mice weighing 20 g were injected with 0.02 mg of Win55-212-2. Mice in the CS and RS groups received an equivalent amount of solvent (a mixture of 10% Tween 80, 10% DMSO, and sterile saline) through intraperitoneal injections on a daily basis (Figure 5A).

During the initial week without Win55-212-2 or solvent treatment, radiation-exposed mice exhibited a significant reduction in center distance and center time in the OFT compared to the control mice (Figures 5B–E; p = 0.0337, Figure 5D; p = 0.0114, Figure 5E), indicating induced anxiety-like behavior. Following Win55-212-2 or solvent treatment Figure 5F, the results showed that, compared with the RS group, the RW group exhibited a significant increase in center distance in the OFT (Figures 5G–J; p = 0.0146, Figure 5I), and increased distance and time in open arms of the EPM on day 14 (Figures 5K–N; p = 0.045, Figure 5L; p = 0.045, Figure 5M). On day 21, compared with the RS group, the RW group displayed increased center distance and time in OFT (Figures 5G–J; p = 0.0406, Figure 5I); p = 0.0297, Figure 5J) and increased distance and time in open arms of EPM (Figures 5K–N; p = 0.0427, Figure 5L; p = 0.0443, Figure 5M). These results demonstrate that dual-frequency EMR caused anxiety-like behaviors in mice, and Win55-212-2 significantly alleviated these behaviors.

Moreover, we evaluated hormone levels in the serum, revealing that the RW group exhibited significantly lower CORT (p = 0.032, Figure 5O) and CRH contents (p = 0.0407, Figure 5P) compared to the RS group. Notably, no significant difference was observed in CORT and CRH content between the RW and CS groups. These findings imply that Win55-212-2 effectively reinstates serum CORT and CRH levels in response to dual-frequency EMR exposure. Subsequently, we assessed the CB1R content (Figures 5Q–R), uncovering that the RW group demonstrated a significant increase (p = 0.0237, Figure 5Q) compared to the RS group, with no notable difference between the RW and CS groups. These results signify that dual-frequency EMR induces downregulation of CB1R in the mouse cerebral cortex, which Win55-212-2 can efficiently ameliorate. In the final phase of our investigation into the effects of Win55-212-2 on endocannabinoid content in the brain cortex and hippocampus, we measured 2-AG and AEA contents (Figures 5S–T). The 2-AG content in the cortex of the RW group was significantly restored compared to the RS group (p = 0.0404, Figure 5S), indicative of Win55-212-2’s effectiveness in alleviating the downregulation of 2-AG induced by dual-frequency EMR.

In conclusion, Win55-212-2 may ultimately improve anxiety-like behavior induced by dual-frequency EMR by restoring HPA axis serum hormone levels and brain cortex 2-AG and CB1R content.

Given that Win55-212-2 acts as an agonist for the cannabinoid receptors CB1 and CB2, we aimed to elucidate the specific involvement of CB1R in anxiety induced by dual-frequency EMR. To assess this, we employed the CB1R antagonist AM251 to ascertain whether the anti-anxiety effects of Win55-212-2 could be inhibited. The RWA group mice received intraperitoneal injections of WinAM251 (Win55-212-2 and AM251 combination, a cannabinoid receptor antagonist, APExBIO Technology LLC, USA) daily, with a dose of 1 mg/kg. In contrast, mice in the CS and RS groups were intraperitoneally injected with an equal amount of solvent (a mixture of 10% Tween 80, 10% DMSO, and sterile saline) daily (Figure 6A).

Upon completion of the WinAM251 or solvent treatment (Figure 6B), the results revealed that, in comparison with the RS group, there were no significant differences in the center distance and center time of the OFT (Figures 5C–F). Similarly, there were no significant differences in the distance and time in the open arms of the EPM (Figures 5G–J) in the RWA group on day 21. We already know that Win55-221-2 can ameliorate anxiety-like behavior induced by dual-frequency EMR. However, following the WinAM251 treatment, the anxiety-like behavior did not improve. These findings indicate that AM251 effectively antagonizes the anti-anxiety effects of Win55-221-2, suggesting a direct association between anxiety-like behaviors induced by dual-frequency EMR in mice and the CB1R of the ECS.

Eight-week-old male C57BL/6 J mice (20.00 ± 0.43 g) were purchased from SPF Biotechnology Co., LTD. (Beijing, China). The mice were housed in standard laboratory conditions, maintaining a 12 h light–dark cycle (7:30 to 19:30) at an ambient temperature of 22 ± 2°C and a relative humidity of 50–60%. The mice ad libitum access to a standard diet and tap water freely. All experimental animal procedures were approved by the Institutional Animal Care and Use Committee of the National Beijing Center for Drug Safety Evaluation and Research.

The radiation equipment used was a general electromagnetic RC constructed by the Tongning Wu team of the Environmental and Security Department of the China Academy of Information and Communication Technology (Li et al., 2016). The RC basic equipment consists of a shielding body, radio frequency (RF) shield door, and RF agitator, and essential equipment includes the signal source and power amplifier. RC is frequently employed to simulate EMR environments and is widely used in electromagnetic biomedicine and conservation research. RC offers notable advantages. On the one hand, it provides effective shielding to prevent external electromagnetic environment interference, eliminating EMR contamination. On the other hand, the electromagnetic environment it produces has a uniform field strength, making it suitable for biological experiments.

RC supports radiative modeling at any frequency between 0.8 and 4 GHz. The experimental parameters are detailed in Table 1, with frequencies set at 2.65 and 0.8 GHz and the electric field intensity calculated based on the mouse’s body weight, ensuring a specific absorption ratio (SAR) of 4 W/kg. According to guidelines from the International Commission on Non-Ionizing Radiation Protection (ICNIRP, 2020), 4 W/kg is considered a safety threshold. This indicates the radiation intensity at which the body core temperature increases by 1°C after 30 min of whole-body exposure (ICNIRP, 2020). The radiation duration of mice was 4 h per day, from 8:00 to 12:00.

During radiation exposure, the rectal temperature of the mice was monitored in real-time using a fiber-optic thermometer (model THR-NC-1084C, FISO, China). The system includes an optical fiber temperature sensor, a rectal intervention catheter, and a mouse fixator. Mice were fixed within a conical cylinder mouse fixator made of acrylic material, with a tail baffle for stability to facilitate the measurement of mouse anus temperature. Placed on an RC-exposed platform, mice were lubricated with a small amount of Vaseline, and a metal-free fiber optic thermometer sensor (2 mm in diameter) was gently inserted into the anus to a depth of approximately 7 mm. The mouse’s anus temperature was measured in real-time, and rectal temperature data was recorded via the optical fiber before, during, and after radiation exposure.

The equipment was purchased from Shanghai Gilang Software Technology Co., LTD (Shanghai, China). OFT was performed in an open plastic box with a size of 50 × 50 × 40 cm³. Mouse trajectories were monitored for 5 min using the animal behavior analysis software. The total distance, center distance, and center time were used as analysis parameters to assess anxiety-like behavior. Prior to the experiment the mouse was placed in the experimental environment for more than 1 h to avoid stress. In addition, quiet was ensured during the trial. The apparatus was cleaned with a 75% ethanol solution before each mouse was introduced.

The equipment was purchased from Anhui Zhenghua Biological Instrument Equipment Co., LTD (Anhui, China). The EPM consisted of two open arms (40 × 5 × 10 cm³) and two closed arms (40 × 5 × 10 cm³) connected to a common central area (5 × 5 cm²). The apparatus was raised to a height of 50 cm from the ground. Mouse trajectories were monitored for 5 min using animal behavior analysis software. The distance and time in open arms, and open arm entries were used as parameters to assess anxiety-like behavior. Prior to the experiment the mouse was placed in the experimental environment for more than 1 h to avoid stress. In addition, quiet was ensured during the trial. The maze was cleaned with a 75% ethanol solution before each mouse was introduced.

The tail suspension test is a classic experimental paradigm for evaluating depression-like behavior. Before initiating the experiment, the tape was wrapped approximately 1 cm from the tail tip of the mice. A clip was used to secure the tape, allowing the head of the mice to hand upside down on the support, with the nose tip kept approximately 15 cm away from the ground. Video equipment automatically recorded the activity of the mice for 6 min, with the first 2 min serving as an adaptation stage and the second 4 min being the test stage. The duration in which the mice gave up struggling and remained still within the 4-min period was statistically analyzed using animal behavior software to evaluate depression-like behavior. Additionally, a quiet environment was ensured during the trial, and the apparatus was cleaned with a 75% ethanol solution before introducing each mouse.

Following the behavioral experiments, mice were anesthetized by intraperitoneal injection of 1% sodium pentobarbital. After achieving complete anesthesia, blood was drawn from the tail vein of the mice. The collected blood was refrigerated at 4°C overnight, and the upper serum was obtained after centrifugation at 4000 rpm for 10 min. The NE and CRH contents were detected using the mouse NE/CRH ELISA kit (Shanghai Enzyme-Linked Biotechnology Co., LTD., China). In contrast, CORT content was seen utilizing the mouse CORT ELISA kit (Enzo, USA). The experimental procedures were conducted in accordance with the provided instructions.

After achieving complete anesthesia, mice were dissected, and the brain cortex and hippocampus were extracted from three independent biological mice in each group for transcriptome analysis. RNA sequencing was performed by Novogene Co., LTD (Beijing, China). Genes with an adjusted p-value <0.05 were considered differentially expressed. A detailed protocol is provided in Supplementary Information.

The brains of six mice randomly sacrificed in each group were swiftly removed. Subsequently, the cerebral cortex and hippocampus of the mice were isolated and snap-frozen in liquid nitrogen, followed by storage at −80°C. The examination of endocannabinoids 2-AG and AEA in the cerebral cortex and hippocampus was conducted using the mouse endocannabinoids 2-AG/AEA ELISA kit (Shanghai FanTai Biotechnology Co., Ltd., China), and the experimental procedures were carried out according to the provided instructions.

The brains of six mice randomly sacrificed in each group were swiftly removed. Subsequently, we isolated the hippocampus, cortex, and thalamus of mice to be snap-frozen in liquid nitrogen and stored at −80°C. These brain tissues were ground at a ratio of 1:20, meaning 1 mg tissue was added to 20 μL RIPA lysate (Beijing Bomaide Biotechnology Co., Ltd., China), and the supernatant was obtained by centrifugation. The total protein concentration of the samples was then adjusted using the BCA kit (Thermo Fisher, USA), and 5 × SDS-PAGE loading buffer was added. The samples were boiled in a water bath at 100°C for 10 min, removed, cooled on ice, and stored at −20°C. Subsequently, gel (Beijing Pulilai Gene Technology Co., Ltd., China), sample loading (2 μg/μL, 6 μL), electrophoresis run (90 V, 20 min; 110 V, one hour), wet membrane transfer (400 mA, 90 min), and PVDF membrane wash in TBST for 5 min were carried out. The PVDF membrane was blocked in 5% skim milk powder for one hour, followed by incubation in the primary antibody (internal reference: GAPDH and TUBULIN; primary antibodies: CB1R, FAAH, and MGL, all at a ratio of 1:1000, Abcam, American) at 4°C overnight. The next day, after one hour of incubation at room temperature, the PVDF membranes were washed with TBST three times, each time for 10 min. Secondary antibodies (Goat anti-rabbit and goat anti-mouse, Beijing Rambolide Trading Co., LTD, China.) were incubated for one hour and washed three times in TBST for 15 min each time. The developer solution (Lablead, China) was prepared with a ratio of liquid A to liquid B of 1:1, and the exposure was conducted using a developer (Tanon, China). Image J- win64 1.51 software was used to quantify the grayscale values of the WB strips.

Statistical analysis was performed using GraphPad Prism 8.4.2 software, and the data are expressed as the mean ± standard error of the mean (M ± SEM). Differences between two groups were analyzed by unpaired t-test, and differences between multiple groups were tested by one-way ANOVA. p-values <0.05 were considered to indicate significant differences.