All animal procedures were approved by the Institutional Animal Care and Use Committee of the Beijing Institute of Radiation Medicine (Protocol No. IACUC-DWZX-2021-685). Ten-week-old wild-type C57BL/6N mice (SPF Biotechnology Co., Ltd, Beijing, China) weighing 22–24 g, with an equal number of males and females, were housed at the Experimental Animal Center of the Beijing Institute of Radiation Medicine. Mice were randomly divided into a control group (Con), static magnetic field exposure group (MF), microwave exposure group (MW), and static magnetic field and microwave exposure group (MF&MW). All mice were housed in an environment maintained at 18–22 °C, relative humidity of 50–60%, and a 12:12 h light-dark cycle, with free access to food and water.

Animals were placed in a transparent, perforated acrylic box which allowed free movement, and their whole body was exposed to EMF. For static magnetic field exposure, mice were placed in a uniform field 1 cm away from the center of the magnetic field, with a field intensity of 100 mT, for 1 h (Fig. 1A). For microwave exposure, mice were placed on a platform, 43 cm away from the transmitter antenna, with a center frequency of 9.375 GHz and an average power density of 12 mW/cm², for 15 min (Fig. 1B). Electromagnetic dosimetry was performed using a finite-difference time-domain algorithm implemented in the Sim4Life platform (ZMT Zurich MedTech AG, Zurich, Switzerland), incorporating a mouse-specific anatomical model (Diggy Male Nude Normal Mouse Model, 23.3 g). Both the exposure antenna and mouse geometry were explicitly modeled to calculate whole-body and spatial SAR distributions under the experimental exposure conditions. Whole-body averaged SAR values ranged from 2.33 to 2.83 W/kg across individual mice (mean, 2.58 W/kg). Under rotation around the platform center during exposure, absorbed dose was considered comparable across animals. Simulated SAR distributions showed no evidence of focal hotspots (Supplementary Fig. 1A). To directly assess potential thermal effects during EMF exposure, rectal temperature of mice was monitored using optical fiber thermometry. Rectal temperature was recorded for 5 min prior to exposure to establish baseline values, continuously monitored throughout the 15-min microwave exposure period, and further measured for 5 min following the end of exposure. No significant elevation in rectal temperature was observed during or after EMF exposure (Supplementary Fig. 1B). The control group mice were placed in the same acrylic box under the same experimental conditions, exposed for the same duration, but without a static magnetic field and microwave exposure. Mice in the MF group were exposed only to the static magnetic field, whereas those in the MW group were exposed only to microwaves. Mice in the combined exposure group were exposed to microwaves after exposure to a static magnetic field.

Fig. 1.

Diagram of electromagnetic field exposure system. (A) Static magnetic field exposure. (B) Microwave exposure.

The conditioned fear memory test was conducted using the Conditioned Fear Memory System (NIR-022; Med Associates Inc., St. Albans, VT, USA). The experiment was divided into two stages: fear memory training (training) and fear memory testing (retrieval-based test). Two sound stimuli with frequencies of 12 and 2 kHz were selected, both with an intensity of 75 dB. During the training phase, mice were placed in the fear chamber for 150 s for acclimation. The two kinds of sound stimuli were presented in a pseudo-random manner, with a duration of 20 s for each stimulus and a 30-s interval between stimuli. Each of the two sound stimuli was presented 10 times. The 12 kHz stimulus was paired with an electric foot shock of 0.6 mA intensity during the last 2 s of sound duration. The 2 kHz stimulus was presented without a paired electric foot shock. Each mouse was taken out of the fear chamber 2 min after the last sound stimulation and the experiment was repeated with the next mouse. The testing phase was designed to assess the retrieval of an already established conditioned fear memory. After a period of 150 s to allow the mouse to adapt to the fear chamber, mice were presented with a 12 kHz tone lasting 100 s without foot shock, followed by removal from the chamber 2 min later. After a 2-h interval, mice were placed in the fear chamber again, and a 2 kHz sound stimulus was administered, lasting 100 s. Freezing behavior during tone presentation was quantified as an index of conditioned fear memory expression. The duration of freezing response for each mouse was measured 1 s after the onset of continuous freezing behavior, and the percentage of freezing time during the sound stimulation was calculated. Freezing was defined as complete immobility, excluding breathing.

The structural organization of the brain underlies its functioning. Previous behavioral test showed a decline in conditioned fear memory at 7 days after EMF exposure. Thus, the morphological changes in the Au1 and BLA brain regions, which are closely related to conditioned fear memory, were observed by transmission electron microscopy at 7 days after EMF exposure.

Mice were anesthetized with 1% pentobarbital sodium (Cat. No. YU021, Shanghai Yuyan Scientific Instruments Co., Ltd., Shanghai, China) (50 mg/kg) via intraperitoneal injection, followed by cardiac perfusion with 4% paraformaldehyde for fixation. The brain tissue was then harvested. For Nissl staining, the brain tissue was rapidly placed in a 4% paraformaldehyde solution (Cat. No. P4500, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) for one-week fixation, and then dehydrated with graded ethanol, cleared in xylene (Cat. No. CAS 1330-20-7, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) and embedded in paraffin. Coronary sections (3-µm thick) of the auditory cortex and amygdala regions were made. The sections were dewaxed with xylene, dehydrated with graded ethanol, and stained with 0.5% toluidine blue solution (Cat. No. BL999A, Biosharp, Hefei, Anhui, China) at room temperature for 25 min. After differentiation with 70% ethanol, xylene transparency, and neutral-resin mounting, the sections were observed using an optical microscopy system (DX12; 3DHISTECN, Budapest, Hungary). Image analysis was performed using ImageJ (2024, Fiji 1.54f) software (NIH, Bethesda, MD, USA) to calculate the mean optical density (MOD) of Nissl bodies.

For ultrastructural morphology, the left Au1 and amygdala were dissected on ice, placed in 2.5% glutaraldehyde (Cat. No. CAS 111-30-8; Xiamen Haibiao Technology Co., Ltd., Xiamen, China) for 2 h, followed by fixation with 1% osmic acid (Cat. No. O109047, Aladdin Bio-Chem Technology Co., Ltd., Shanghai, China) for 2 h, dehydrated with graded ethanol and acetone (Cat. No. A475456, Aladdin Bio-Chem Technology Co., Ltd., Shanghai, China), embedded in Epon 812, and after positioning the semi-thin sections, ultra-thin sections were prepared. Then double staining with uranium acetate and lead citrate was performed, and the ultrastructural morphologies were observed using a transmission electron microscope (HT-7800; Hitachi, Tokyo, Japan).

For influence observation, the frozen sections of brain were prepared using a cryostat (CM1950; Leica, Wetzlar, Germany) with a thickness of 20 µm. After mounted with an anti-fluorescence quenching mounting medium (containing DAPI) (104139, Abcam, Cambridge, UK), the sections were observed using a laser scanning confocal microscope (LSM880; Leica, Germany). Quantitative analysis of the fluorescent images was performed using ImageJ software (Fiji, version 1.54f, NIH, Bethesda, MD, USA), and the relative fluorescent area (area %) was calculated.

The viruses used included rAAV-hSyn-EGFP (Cat. No. BC-0020, BrainCase Biotechnology Co., Ltd., Shenzhen, Guangdong, China), rAAV-hSyn-mCherry (Cat. No. BC-0023, BrainCase Biotechnology Co., Ltd.), rAAV-hSyn-jGCaMP7b (Cat. No. BC-0258, BrainCase Biotechnology Co., Ltd.), rAAV-CaMKII$\alpha$-hM3D(Gq)-EGFP (Cat. No. BC-0139, BrainCase Biotechnology Co., Ltd.), and rAAV-CaMKII$\alpha$-hM4D(Gi)-EGFP (Cat. No. BC-0245, BrainCase Biotechnology Co., Ltd.). Mice were anesthetized by inhalation of 4–5% isoflurane (Cat. No. R510-22-2, Ruiwode Life Science Co., Ltd.) and fixed in a stereotaxic apparatus (Cat. No. 68025, Ruiwode Life Science Co., Ltd.). Erythromycin ointment (Cat. No. H36020018, Jiangxi Decheng Pharmaceutical Co., Ltd., Nanchang, Jiangxi, China) was applied for eye protection, and a heating pad was used to maintain body temperature at 37 °C. After disinfection, the scalp was removed to expose the skull. A mini drill was used to drill holes into the skull, and the virus was injected into the target brain region at a rate of 30 nL/min via a glass micro-needle using a micro-syringe pump. The coordinates for Au1 and BLA were determined based on the mouse brain atlas: Au1 (anterior-posterior (AP): –2.3 mm, medial-lateral (ML): $\pm$4.3 mm, dorsal-ventral (DV): –3.2 mm) and BLA (AP: –1.2 mm, ML: $\pm$3.8 mm, DV: –4.9 mm). After injection, the needle was left for 10 min. Following removal of the needle, dental cement was applied. The mouse was then allowed to awaken naturally before being returned to its housing cage for recovery.

To detect calcium activity in Au1 and BLA neurons, rAAV-hSyn-jGCaMP7b was injected into the Au1 and BLA of mice and ceramic inserts were implanted at the virus injection sites. To detect Au1-BLA neuronal fiber projections, rAAV-hSyn-EGFP was injected into the Au1, and rAAV-hSyn-mCherry was injected into the BLA. To regulate the activity of Au1 glutamatergic neurons, rAAV-CaMKII$\alpha$-hM3D(Gq)-EGFP or rAAV-CaMKII$\alpha$-hM4D(Gi)-EGFP was injected into the Au1.

Three weeks after AAV-jGCaMP7b injection into the Au1 and BLA, the conditioned fear memory test was conducted. A fiber-optic recording system (QAXK-FPS-SS-MC-LED, Thinker Tech Nanjing Bioscience Inc., Nanjing, China) was used to continuously record neuronal calcium activity throughout the entire behavioral session. The fiber-optic patch was inserted through a small hole above the fear chamber and connected to the ceramic implant in the mouse. A 473-nm laser was delivered through the optical fiber to excite the genetically encoded calcium indicator jGCaMP7b, and fluorescence signals were continuously acquired during the entire sound stimulation period, including both freezing and non-freezing behavioral states.

To characterize neural activity associated with fear expression, freezing onset was identified based on frame-by-frame behavioral scoring. For freezing-aligned analyses, calcium traces were aligned to the onset of each freezing episode. The local trough immediately preceding the rise of the calcium signal waveform was defined as the reference time point (T = 0 s) for visualization of freezing-associated calcium dynamics. Calcium signals were expressed as $\mathrm{\Delta }$F/F₀ = (F_signal – F₀)/F₀, where F₀ was calculated as the mean fluorescence value during the –2 to 0 s baseline period preceding T = 0 s. In addition, to assess stimulus-evoked calcium responses independent of behavioral state, a separate tone-onset-locked analysis was performed. In this analysis, time zero was defined as the onset of the auditory stimulus (100 s sound onset), and calcium signals were extracted within a –2 to +10 s window relative to tone onset for both the 12 kHz conditioned tone and the 2 kHz non-conditioned tone. All trials were included regardless of subsequent freezing behavior. Data analysis was performed using MATLAB (Version R2024a, MathWorks, Natick, MA, USA).

Three weeks after injection of AAV-CaMKII$\alpha$-hM3D or AAV-CaMKII$\alpha$-hM4D, mice received an intraperitoneal injection of either saline (NS) or clozapine-N-oxide (Cat. No. CNO-25 mg, BrainCase Biotechnology Co., Ltd., Shenzhen, Guangdong, China), prepared at 0.33 mg/mL and administered at a volume of 0.01 mL/g body weight (approximately 3.3 mg/kg), 30 min prior to behavioral testing. CNO administration was used to modulate the activity of Au1 CaMKII$\alpha$-expressing excitatory neurons expressing designer receptors exclusively activated by designer drugs, whereas saline served as a vehicle control. Following these manipulations, conditioned fear behavioral tests were conducted, and freezing behavior during sound stimulation was quantified to assess changes in fear retrieval.

Double immunofluorescence staining was used to detect the expression of neuronal-specific markers (ChAT and GAD 67) and c-Fos was used as a neuronal activity marker. Briefly, frozen sections of brain tissue were treated with 0.25% Triton X-100 for 25 min, followed by antibody blocking with 1% bovine serum albumin (BSA) (Cat. No. A8020, Solarbio Science & Technology Co., Ltd., Beijing, China) for 2 h. Then, the mixture diluent of c-Fos antibody (1:1000, Cat. No. 2250, Cell Signaling Technology, Danvers, MA, USA) and GAD 67 (1:1000, Cat. No. ab26116, Abcam, Cambridge, UK) antibody, or c-Fos antibody (1:200, Cat. No. ab208942, Abcam) and ChAT (1:1000, Cat. No. ab6168, Abcam) antibody diluent was added and the sections were incubated at 4 °C overnight. After washing three times with 0.025% Triton X-100 (Cat. No. G3068, Wuhan Servicebio Technology Co., Ltd., Wuhan, Hubei, China) in phosphate-buffered saline (PBS) (Cat. No. G4202, Wuhan Servicebio Technology Co., Ltd.), a mixture of goat anti-rabbit (1:500, Cat. No. Alexa Fluor® 488, Abcam) and goat anti-mouse (1:500, Cat. No. Alexa Fluor® 594, Abcam) secondary antibody was added and the sections were incubated at room temperature for 2 h, followed by washing with 0.025% Triton X-100 in PBS.

After mounting with anti-fluorescence quencher medium, the sections were observed using a laser scanning confocal microscope (LSM880; Leica, Wetzlar, Germany). Quantitative analyses were performed with ImageJ software. For each section, regions of interest (ROIs) corresponding to the BLA were manually delineated according to anatomical landmarks. Following background subtraction and thresholding with identical parameters across groups, the fluorescence-positive areas for c-Fos and for ChAT or GAD67 were measured independently. To estimate relative neuronal activity within specific cell populations, cholinergic and GABAergic activity indices were calculated as the ratio of c-Fos-positive area to ChAT-positive area (Area-c-Fos/Area-ChAT) or to GAD67-positive area (Area-c-Fos/Area-GAD67), respectively. In parallel, fluorescence colocalization was assessed using the Colocalization Finder plugin in ImageJ. Pearson’s correlation coefficients were calculated to quantify the spatial association between c-Fos and ChAT signals, or between c-Fos and GAD67 signals, within the defined ROIs.

All experimental data were reported as the mean and standard error of the mean (SEM). Statistical analysis was performed using SPSS 27.0 software (IBM Corp., Armonk, NY, USA). Core body temperature measurements obtained before, during, and after EMF exposure were analyzed using paired two-tailed Student’s t-tests. The conditioned fear memory behavior was analyzed using two-way repeated-measures ANOVA. For sex-stratified analyses and DPOAE amplitude data were additionally analyzed using multivariate ANOVA with sex and exposure condition as between-subject factors. Quantification data of Nissl staining and immunofluorescence staining were analyzed using two-way ANOVA. The statistical significance of differences in ABR thresholds, neural-fiber projection, and neuronal calcium signals between groups was assessed using independent samples t-tests. p values 0.05 were considered statistically significant.

Fear memory is one of the most important emotions in both humans and animals, and is generally associated with cues such as sound or light stimuli [23]. This study used Pavlov’s classical conditioning model to associate auditory stimuli with foot shock, testing the discrimination of two sounds with different frequencies, and the freezing response in mice. A behavioral paradigm of fear discrimination during retrieval-based testing was established for EMF exposure as shown in Fig. 2A,B.

Fig. 2.

Electromagnetic field exposure attenuates conditioned fear memory retrieval in mice. (A) Timeline of conditioned fear memory experiment. (B) Schematic diagram of the auditory fear conditioning and retrieval-based testing procedure. (C) Percentage of freezing time in mice for each group during the conditioned fear memory training and retrieval-based test phases. (D) Percentage of freezing time response to 2 kHz sound stimulus 6 h after exposure. (E,F) Percentage of freezing time response to 12 kHz (E) and 2 kHz (F) sound stimulus 7 days after exposure. Mean $\pm$ SEM, n = 8, two-way repeated-measures ANOVA. *p 0.05, **p 0.01, ***p 0.001. Con, control group; MF, static magnetic field exposure group; MF&MW, combined static magnetic field and microwave exposure group; MW, microwave exposure group.

During the conditioned fear training phase, freezing behavior progressively increased and exceeded 80% from the sixth to tenth tone presentation, indicating successful acquisition of conditioned fear memory prior to EMF exposure (Fig. 2C). During the test phase, behavioral responses were evaluated at multiple post-exposure time points to assess the retrieval of conditioned fear memory. At 6 h after exposure, the percentage of freezing time response to 2 kHz stimulus was lower in the MF&MW group than in the MF group 6 h (Fig. 2D). The percentage of freezing time in response to 12 kHz stimulus with electric shock 7 d after EMF exposure was significantly lower in the MF&MW group than in the other groups (Fig. 2E). The percentage of freezing time in response to a 2 kHz stimulus without electric shock was also significantly higher in the MF&MW group than in the other groups (Fig. 2F). The ratio of 2:12 kHz freezing time percentage, percentage of freezing time of 12 kHz after 6 h, and the freezing time percentage 14 days after exposure did not differ significantly between groups (Supplementary Fig. 2). These results indicate that combined exposure was associated with reduced retrieval of conditioned fear in mice, with the difference being most apparent at 7 days after exposure, whereas exposure to MF or MW alone did not result in detectable changes in fear-related behavior under the present conditions.

To evaluate whether the EMF-associated alterations in conditioned freezing behavior could be attributed to peripheral auditory dysfunction, auditory function was assessed using ABR and DPOAE. ABR testing revealed no significant differences in auditory thresholds among Con, MF, MW, and MF&MW group at 7 days post-exposure, in either the left or right ear (Fig. 3A,B). These results indicate that EMF exposure did not induce detectable impairments in auditory nerve or brainstem function. In line with these findings, DPOAE recordings revealed no significant group differences in distortion product amplitudes across the tested frequency range in either ear (Fig. 3C,D), suggesting preserved outer hair cell function following EMF exposure. Together, these findings indicate that the attenuation of conditioned freezing behavior observed following EMF exposure is unlikely to result from gross auditory dysfunction and instead supports an interpretation based on altered conditioned fear retrieval rather than primary sensory deficits.

Fig. 3.

Changes of ABR thresholds and DPOAE assessments in mice after electromagnetic field exposure. (A,B) ABR thresholds measured in the left (A) and right (B) ears of mice in all experimental groups at multiple time points (Pre, Post 0 d, Post 7 d, and Post 14 d) after exposure. (C,D) Changes in DPOAE amplitudes in the left (C) and right (D) ear of mice from each experimental group at 7 days after exposure. Mean $\pm$ SEM, n = 8, ABR: t-test; DPOAE: multivariate ANOVA.

Because sex differences have been reported in fear related behaviors [24, 25], we next examined whether EMF exposure induced changes in freezing responses differed between male and female mice. The results showed a significant effect of exposure condition on freezing responses at 7 days after EMF exposure, whereas no sex-related differences or sex-by-exposure interactions were observed for freezing responses to the 12 kHz conditioned stimulus, the 2 kHz non-conditioned stimulus, or the 2:12 kHz freezing ratio (Supplementary Fig. 3D–F). In contrast, no significant effects of sex or sex-by-exposure interactions were detected at 6 h (Supplementary Fig. 3A–C) or 14 days (Supplementary Fig. 3G–I) after EMF exposure for any of the measured fear-related indices. These results indicate that the EMF-associated attenuation of conditioned fear retrieval observed at the 7-day time point was comparable between male and female mice under the present experimental conditions.

Ultrastructure observation 7 days after EMF exposure revealed irregular nuclei, nuclear chromatin condensation and margination, indistinct nuclear membrane, focal vacuolization of swollen mitochondria, and dilation and degranulation of endoplasmic reticulum in some Au1 and BLA neurons. Moreover, blurring of the synaptic clefts, loss of definition of the double membranes, and accumulation of synaptic vesicles in the presynaptic terminal were observed in the EMF exposure groups (Fig. 4A,B). These pathological changes of ultrastructure were more marked in the MF&MW group than in the MF and MW groups, and the changes were more marked in Au1 neurons than in BLA neurons. This suggests that EMF exposure can cause pathological changes of ultrastructure in neurons in the Au1 and BLA regions of mice.

Fig. 4.

Pathological changes of the Au1 and BLA neurons in mice exposed to electromagnetic radiation. (A,B) Ultrastructure of neurons and synapses in the Au1 (A) and BLA (B), Black arrows indicate representative pathological features, including cytoplasmic condensation along the nuclear membrane, mitochondrial swelling, accumulation of synaptic vesicles, and blurred synaptic clefts. (C) Representative images of Nissl body staining in Au1 and BLA. (D,E) Quantification of Nissl body in the Au1 (D) and BLA (E). Mean $\pm$ SEM, n = 6, Two-way ANOVA. *p 0.05. Au1, primary auditory cortex; BLA, basolateral amygdala; Con, control group; MF, static magnetic field exposure group; MF&MW, combined static magnetic field and microwave exposure group; MOD, mean optical density; MW, microwave exposure group. Scale bars: Neurons, 2 µm; Synapses, 500 nm (A,B); 20 µm (C).

The effects of EMF exposure on neurons in the Au1 and BLA regions, were confirmed by investigating the changes of Nissl body in neurons by Nissl staining for both Au1 and BLA 7 days after EMF exposure (Fig. 4C). This revealed a significant reduction in the MOD of Nissl bodies in the Au1 neurons in the MF&MW group, compared with those in the Con group, whereas the MOD of Nissl bodies in the Au1 neurons in the MF and MW groups did not differ significantly from that of the Con group (Fig. 4D). The MOD of Nissl bodies in the BLA neurons in the MF&MW, MF, and MW groups also did not differ significantly from that of the Con group (Fig. 4E). The reduction in the MOD of Nissl bodies indicates a decrease in the amount of rough endoplasmic reticulum and free ribosomes within neurons, suggesting that EMF exposure may inhibit the protein synthesis in Au1 neurons. Additionally, the results of Nissl body staining suggest that the effects of EMF exposure on Au1 neurons were more pronounced than those on the BLA neurons, and that the combined exposure of MF&MW had a more marked effect on Au1 neurons than either MF or MW exposure alone.

Au1 is responsible for processing auditory stimuli related to emotions and transmitting the information to the BLA [26, 27]. Therefore, this study investigated the effects of EMF exposure on the Au1-BLA neural projections by anterograde and retrograde tracing. For anterograde tracing, rAAV2/9-hSyn-EGFP-WPRE-hGHpA was injected into the Au1. For retrograde tracing, rAAV2/retro-hSyn-mCherry-WPRE-hGHpA was injected into the BLA.

The accuracy of the micro-injection was confirmed by the expression of fluorescent protein EGFP or mCherry (Fig. 5A,B). The quantification of Au1-BLA projection (Fig. 5C,D) showed that 7 days after exposure, both the anterograde (Fig. 5E) and retrograde (Fig. 5F) projections were significantly reduced in the MF&MW group compared with the Con group. These findings indicate that EMF exposure is associated with a reduction in projection-consistent labeling along the Au1-BLA pathway, suggesting an altered functional state of the Au1-BLA circuit under EMF exposure. Rather than indicating a definitive loss of anatomical connectivity, this change is more likely to reflect alterations in activity-dependent labeling efficiency or functional properties of Au1-BLA projections.

Fig. 5.

Electromagnetic field exposure is associated with altered Au1-BLA projection intensity. (A,B) Confirmation of the target region Au1 (A) and BLA (B) for micro-injection of rAAV-hSyn-EGFP and rAAV-hSyn-mCherry, respectively. (C,D) Representative images of BLA (C) and Au1 (D) for the Au1-BLA projections by AAV-based tracing. (E,F) Quantification analysis of the Au1-BLA projections by anterograde (E) and retrograde (F) AAV-based tracing. Mean $\pm$ SEM, n = 6, independent samples t-test. *p 0.05, ***p 0.001. AAV, adeno-associated virus. Scale bars: 200 µm (A, B, and low-magnification images in C and D); 50 µm (high-magnification images in C and D).

Au1 is responsible for the integration and transmission of auditory information [12], whereas BLA processes and stores emotional information and coordinates behavior [9, 28]. Both Au1 and BLA play an important role in conditioned fear memory [29]. To investigate the neural mechanisms underlying the attenuation of conditioned fear expression following combined exposure to static magnetic fields and microwaves, we used genetically encoded calcium imaging technology, rAAV-hSyn-jGCaMP7b were injected into the Au1 and BLA brain regions (Fig. 6A). Using fiber-optic recordings combined with conditioned fear memory behavioral tests, we examined calcium activity in Au1 and BLA neurons during the retrieval phase of conditioned fear memory. Calcium signals were first analyzed time-locked to the onset of auditory stimulation, independent of behavioral state. We then analyzed calcium activity in Au1 and BLA neurons during sound-evoked freezing episodes in the conditioned fear memory test, focusing on neural activity associated with fear expression (Fig. 6B).

Fig. 6.

Reduced calcium activity in Au1 and BLA neurons is associated with impaired conditioned fear memory retrieval in mice exposed to electromagnetic fields. (A) Schematic diagram of rAAV-GCaMP injection into the Au1 and BLA brain regions. (B) Schematic diagram of fiber optic recording combined with conditioned fear memory tests. (C–F) Plot of AUC, $\mathrm{\Delta }$F/F₀ (%), and heat maps of calcium signal in the Au1 (C,D) and BLA (E,F) during conditioned fear retrieval elicited by the 12 kHz tone. (G–J) Plot of AUC, $\mathrm{\Delta }$F/F₀ (%), and heat maps of calcium signal in the Au1 (G,H) and BLA (I,J) during conditioned fear retrieval elicited by the 2 kHz tone. Mean $\pm$ SEM, n = 6, independent samples t-test. *p 0.05, **p 0.01. AUC, area under the curve; Con, control group; F, fluorescence signal value; F₀, baseline fluorescence signal value.

Seven days after EMF exposure, tone-onset–aligned analysis revealed no significant differences in stimulus-evoked calcium responses between the MF&MW and control groups, either during presentation of the fear-paired 12 kHz tone or the non-paired 2 kHz tone (Supplementary Fig. 4), indicating that EMF exposure did not alter the immediate neural response to auditory cue onset. In contrast, when calcium signals were analyzed during freezing episodes elicited by the fear-paired 12 kHz tone, a significant reduction in calcium activity was observed in both Au1 and BLA neurons in the MF&MW group compared with controls (Fig. 6C–F). This reduction was not observed during freezing induced by the non-paired 2 kHz tone in the absence of electric shock, where calcium activity in Au1 and BLA neurons did not differ significantly between groups (Fig. 6G–J). Furthermore, calcium activity in Au1 and BLA neurons remained unchanged at 6 h after EMF exposure regardless of whether 12 kHz or 2 kHz auditory stimuli were presented (Supplementary Fig. 5).

These results indicate that EMF exposure does not affect stimulus-locked auditory encoding in the Au1–BLA circuit but selectively reduces circuit engagement during the expression of conditioned fear. The observed decrease in calcium activity is therefore associated with altered state-dependent recruitment of the Au1–BLA circuit during fear expression, rather than disruption of initial sensory processing or memory acquisition.

Glutamate is a neurotransmitter involved in memory encoding [30, 31] and is closely associated with fear memory [23]. Isler et al. [32] found that patients with tinnitus had reduced glutamate levels in the Au1 brain region. Nomura et al. [33] used optogenetic technology to inhibit glutamatergic neurons in the Au1, resulting in the impairment of fear memory and memory retrieval in mice. These findings suggest that glutamatergic neurons in the Au1 may play a critical role in conditioned fear memory. This study used chemogenetic manipulation combined with conditioned fear behavioral tests to examine the functional involvement of Au1 excitatory neurons in EMF-associated changes in conditioned fear retrieval. AAV-CaMKII-hM3D(Gq)/hM4D(Gi) containing the glutamatergic neuron-specific promoter CaMKII was injected into the Au1, and clozapine-N-oxide (CNO) was administered via intraperitoneal injection to selectively regulate glutamatergic neurons in the Au1. The “ON” condition refers to chemogenetic activation of Au1 CaMKII-expressing neurons, whereas the “OFF” condition refers to chemogenetic inhibition of the same neuronal population.

The changes in percentage of freezing time during the conditioned fear memory test phase in mice following EMF exposure were analyzed (Fig. 7A). Seven days after EMF exposure, the percentage of freezing time in the MF&MW-NS group was significantly lower than that in the Con-NS group and Con-ON group when paired with a 12 kHz sound stimulus (Fig. 7B,E), whereas it was higher in the MF&MW-NS group than in the Con-NS and Con-ON groups when paired with a 2 kHz sound stimulus (Fig. 7C,F). The ratio of 2:12 kHz freezing time also increased after EMF exposure (Fig. 7D,G). These results were consistent with those of the previous conditioned fear memory behavioral test, and confirmed that EMF exposure can impair conditioned fear memory in mice.

Fig. 7.

Chemogenetic modulation of Au1 CaMKII-expressing neurons alters conditioned fear memory retrieval in mice following electromagnetic field exposure. (A) Flowchart of the chemogenetic-based conditioned fear memory experiment. (B–D) Following chemogenetic activation of Au1 CaMKII-expressing neurons, the percentage of freezing time response to 12 kHz (B) and 2 kHz (C) sound stimulus, and the ratio of 2:12 kHz (D) in mice. (E–G) Following chemogenetic inhibition of Au1 CaMKII-expressing neurons, the percentage of freezing time response to 12 kHz (E) and 2 kHz (F) sound stimulus, and the ratio of 2:12 kHz (G) in mice. Mean $\pm$ SEM, n = 8. Two-way ANOVA. *p 0.05, **p 0.01, ***p 0.001. Con, control group; MF, static magnetic field exposure group; MF&MW, combined static magnetic field and microwave exposure group; MW, microwave exposure group; NS, normal saline; ON, chemogenetic activation of Au1 CaMKII-expressing neurons; OFF, chemogenetic inhibition of Au1 CaMKII-expressing neurons.

Notably, following chemogenetic activation of Au1 glutamatergic neurons, the percentage of freezing time in the MF&MW-ON group was significantly higher than that in the MF&MW-NS group mice when exposed to 12 kHz stimulus (Fig. 7B), whereas the percentage of freezing time in the MF&MW-ON group mice was significantly lower than that in the MF&MW-NS group mice when exposed to a 2 kHz stimulus (Fig. 7C), and the ratio of 2:12 kHz freezing time was significantly lower in the MF&MW-ON group than in the MF&MW-NS group (Fig. 7D).

However, following inhibition of Au1 glutamatergic neurons, the percentage of freezing time in the MF&MW-NS, MF&MW-OFF, and Con-OFF groups was significantly lower than that in the Con-NS group when exposed to 12 kHz stimulus (Fig. 7E), whereas the percentage of freezing time (Fig. 7F) and the ratios of 2:12 kHz freezing time (Fig. 7G) ratio were significantly higher in the MF&MW-NS, MF&MW-OFF, and Con-OFF groups than in the Con-NS group when exposed to 2 kHz stimulus.

These results indicate that chemogenetic modulation of Au1 CaMKII-expressing neurons is associated with bidirectional changes in conditioned fear expression following EMF exposure. Together, these findings support the functional involvement of Au1 CaMKII-expressing neurons in the modulation of conditioned fear expression under EMF exposure conditions, rather than serving as definitive evidence of cell-type exclusive causality.

To further examine the functional relationship between Au1 neuronal activity and BLA dynamics during conditioned fear retrieval, we combined chemogenetic manipulation with genetically encoded calcium imaging to assess changes in BLA neuronal calcium activity following EMF exposure. Seven days after EMF exposure, during presentation of the 12 kHz sound triggered mouse freezing behavior, calcium activity in BLA neurons was significantly higher in the MF&MW-ON group than in the MF&MW-NS group (Fig. 8A,B). In contrast, during presentation of the 2 kHz tone, no significant difference in BLA calcium activity was observed between the MF&MW-ON and MF&MW-NS groups (Fig. 8C,D). These findings indicate that chemogenetic activation of Au1 CaMKII-expressing neurons is associated with a partial restoration of BLA neuronal calcium activity following EMF exposure, supporting a functional link between Au1 activity and BLA engagement during conditioned fear retrieval. After EMF exposure, activation of Au1 glutamatergic neurons in the brain region only increased the calcium activity in BLA neurons after 12 kHZ sound signals, but not after 2 kHz sound signals, possibly because the microcircuits in the BLA mainly encode aversive stimuli [34, 35, 36]. These findings suggest that coordinated activity between Au1 and BLA is associated with modulation of conditioned fear memory retrieval following EMF exposure.

Fig. 8.

Activation of Au1 CaMKII-expressing neurons is associated with increased calcium activity in BLA neurons after electromagnetic field exposure. (A,B) Under EMF exposure, following chemogenetic activation of Au1 CaMKII-expressing neurons, AUC analysis (A), mean $\mathrm{\Delta }$F/F₀ traces and corresponding heat maps (B) of calcium signals in the BLA during presentation of the 12 kHz conditioned tone. (C,D) Under EMF exposure, AUC analysis (C), mean $\mathrm{\Delta }$F/F₀ traces and corresponding heat maps (D) of calcium signals in the BLA during presentation of the 2 kHz tone. Mean $\pm$ SEM, N = 6, independent samples t-test. *p 0.05. Au1, primary auditory cortex; AUC, area under the curve; BLA, basolateral amygdala; MF&MW, combined static magnetic field and microwave exposure group; NS, normal saline; ON, chemogenetic activation of Au1 glutamatergic neurons.

A neural network consists of intricate neurons interconnected via synapses. Glutamatergic neurons are the primary presynaptic neurons, and Au1 glutamatergic neurons can serve as auditory input neurons. In addition, GABAergic neurons and cholinergic neurons are involved in the acquisition and expression of memory. To explore potential downstream neuronal populations associated with Au1-BLA circuit modulation following EMF exposure, neuronal activity in cholinergic and GABAergic neurons within the BLA was examined using immunofluorescence labeling of c-Fos in combination with cell-type–specific markers (ChAT and GAD67) (Fig. 9A–D).

Fig. 9.

Changes in cholinergic and GABAergic neuronal activity in the BLA associated with activation of Au1 CaMKII-expressing neurons after electromagnetic field exposure. (A,B) Representative images of ChAT and c-Fos expression in the BLA of mice in the Con-NS and Con-ON groups (A), and the MF&MW-NS and MF&MW-ON groups (B). (C,D) Representative images of GAD 67 and c-Fos expression in the BLA of mice in the Con-NS and Con-ON groups (C), and the MF&MW-NS and MF&MW-ON groups (D). (E) Quantification of cholinergic activity index, calculated as the ratio of c-Fos-positive area to ChAT-positive area (Area-c-Fos/Area-ChAT). (F) Quantification of the Pearson’s correlation coefficient reflecting the spatial association between ChAT-positive area and c-Fos-positive area. (G) Quantification of GABAergic activity index, calculated as the ratio of c-Fos-positive area to GAD67-positive area (Area-c-Fos/Area-GAD67). (H) Quantification of the Pearson’s correlation coefficient reflecting the spatial association between GAD67-positive area and c-Fos-positive area in the BLA. Mean $\pm$ SEM, n = 6, Two-way ANOVA. *p 0.05, ***p 0.001. White arrows indicate cells positive for the respective markers (c-Fos, ChAT, or GAD67). Scale bars: 20 µm (A–D).

Quantitative immunofluorescence analysis revealed that, 7 days after EMF exposure, the MF&MW-NS group exhibited both a significantly increased cholinergic activity index—calculated as the ratio of c-Fos–positive area to ChAT-positive area—and an enhanced spatial correlation between ChAT and c-Fos expression in the BLA, compared with the Con-NS and MF&MW-ON groups (Fig. 9E,F). These findings indicate elevated activation and tighter functional coupling of cholinergic neurons in the BLA following EMF exposure. Notably, chemogenetic activation of Au1 CaMKII-expressing neurons significantly attenuated both the increase in cholinergic activity and the ChAT–c-Fos correlation, suggesting that Au1 CaMKII-expressing neuronal activation modulates EMF-associated alterations in BLA cholinergic neuronal activity.

By contrast, the activity index derived from GAD67-positive neurons did not differ significantly among groups (Fig. 9G,H). This suggests that GABAergic neuronal activity in the BLA was not detectably altered by EMF exposure or by Au1 CaMKII-expressing neuronal activation under the present experimental conditions.