Long-term radiofrequency electromagnetic fields exposure attenuates cognitive dysfunction in 5×FAD mice by regulating microglial function

Abstract

We have previously found that long-term effects of exposure to radiofrequency electromagnetic fields in 5×FAD mice with severe late-stage Alzheimer’s disease reduced both amyloid-β deposition and glial activation, including microglia. To examine whether this therapeutic effect is due to the regulation of activated microglia, we analyzed microglial gene expression profiles and the existence of microglia in the brain in this study. 5×FAD mice at the age of 1.5 months were assigned to sham- and radiofrequency electromagnetic fields-exposed groups and then animals were exposed to 1950 MHz radiofrequency electromagnetic fields at a specific absorption rate of 5 W/kg for 2 hours/day and 5 days/week for 6 months. We conducted behavioral tests including the object recognition and Y-maze tests and molecular and histopathological analysis of amyloid precursor protein/amyloid-beta metabolism in brain tissue. We confirmed that radiofrequency electromagnetic field exposure for 6 months ameliorated cognitive impairment and amyloid-β deposition. The expression levels of Iba1 (pan-microglial marker) and colony-stimulating factor 1 receptor (CSF1R; regulates microglial proliferation) in the hippocampus in 5×FAD mice treated with radiofrequency electromagnetic fields were significantly reduced compared with those of the sham-exposed group. Subsequently, we analyzed the expression levels of genes related to microgliosis and microglial function in the radiofrequency electromagnetic fields-exposed group compared to those of a CSF1R inhibitor (PLX3397)-treated group. Both radiofrequency electromagnetic fields and PLX3397 suppressed the levels of genes related to microgliosis (Csf1r, CD68, and Ccl6) and pro-inflammatory cytokine interleukin-1β. Notably, the expression levels of genes related to microglial function, including Trem2, Fcgr1a, Ctss, and Spi1, were decreased after long-term radiofrequency electromagnetic field exposure, which was also observed in response to microglial suppression by PLX3397. These results showed that radiofrequency electromagnetic fields ameliorated amyloid-β pathology and cognitive impairment by suppressing amyloid-β deposition-induced microgliosis and their key regulator, CSF1R.

Introduction

Alzheimer’s disease (AD), which is characterized by abnormal accumulation of extracellular senile (amyloid) plaques and neurofibrillary tangles, is a progressive neurodegenerative disease that leads to progressive cognitive dysfunction. Neuronal loss is found in the brain regions of AD patients, particularly in the neocortex and hippocampus, which might be related to the clinical manifestation of AD (Donev et al., 2009). While the accumulation of amyloid-β (Aβ) plaques is considered to be a hallmark of the disease (Selkoe, 2001), its pathogenesis is not yet fully understood. However, preclinical and clinical studies have recently revealed that the innate immune system is involved in AD pathogenesis (Heneka et al., 2014) and that neuroinflammation that is associated with AD is driven by the intrinsic immune cells of the brain and worsens with disease progression (Heppner et al., 2015).

Microglia, which constitute 5–10% of all brain cells, are a key element of the central nervous system that are responsible for normal brain functioning and mediate tissue homeostasis and neuroinflammation (Frost and Schafer, 2016; Li and Barres, 2018). Microglia can remove dead and dying cells by phagocytosis, which plays an important role in the maintenance of brain functions (Galloway et al., 2019). Although the initial immune mechanisms are described as beneficial in major neurodegenerative diseases, sustained activation of microglia can cause further neuronal dysfunction due to the secretion of proinflammatory mediators and neurotoxic factors (Heneka et al., 2014), indicating that chronically activated microglia might aggravate AD pathology. Because microglia have been implicated in the etiology of AD progression, targeting microglia is promising for the treatment of AD (Wes et al., 2016). A previous study demonstrated that microglial deletion by a colony-stimulating factor 1 receptor (CSF1R) inhibitor impaired Aβ plaque formation and reversed hippocampal gene expression in mice model of AD, indicating the beneficial effect of microglial deletion on AD progression (Spangenberg et al., 2019). However, a consensus on the role of microglia in the progression of AD is lacking.

Previous studies have attempted to elucidate the ultimate etiology of AD (Blennow et al., 2006; Nelson et al., 2009; Liu et al., 2019), and no therapeutic drugs have been successful, indicating that novel therapeutic strategies are required. Recently, as the potential use of nonpharmacological interventions to prevent or treat AD has gained attention, the biological consequences of nonpharmacological therapies have received wide consideration. Abnormal behavior was shown to be reduced by electroconvulsive stimulation in the 5×FAD mouse model of AD (Svensson et al., 2021). Previously, we reported that exposure to RF-EMFs for 8 months might have a preventive effect on Aβ-like pathology in 5×FAD mice (Jeong et al., 2015). Long-term RF-EMFs exposure significantly inhibited Aβ deposition in the brain and suppressed the expression of ionized calcium-binding adapter molecule 1 (Iba1), a microglial marker. These effects were found more dominantly in the hippocampus than in the entorhinal cortex or the whole brain (Jeong et al., 2015). We also reported that long-term RF-EMFs increased hippocampal glucose uptake and improved hippocampal-dependent memory behaviors in novel object recognition tests in 5×FAD mice compared with the sham exposure group (Son et al., 2018). Therefore, to clarify the therapeutic mechanism of RF-EMFs on Aβ-like pathology in 5×FAD mice, we focused on the neurobiological changes of RF-EMFs in the hippocampal region of the brain.

In this study, we examined the effects of 6-month-RF-EMFs exposure on AD pathology and cognition in 5×FAD mice. Additionally, changes in the expression levels of genes related to neuroinflammation (IL-1β, IL-4, IL-6, Csf1r, CD68, and Ccl6) and microglial function (Trem2, Fcgr1, Ctss, and Spi1) were examined in the mouse hippocampus. Furthermore, we aimed to identify the role of neuroinflammation after exposure to RF-EMFs by comparing the brains of 5×FAD mice with those of mice with microglial suppression.

Methods

Animals and drug administration

Female 5×FAD mice carrying three human amyloid precursor protein (APP) genes (the Swedish [K670N/M671L], Florida [I716V], and London [V717I] mutations), as well as two human presenilin1 genes (M146L and L286V) were utilized (RRID: MMRRC_034848-JAX). Since it has been reported that expression changes in inflammatory mediators and glial markers are more robust in female mice than male 5×FAD mice (Manji et al., 2019), we used female 5×FAD mice for this study. We used age-matched female B6/SJL mice as the wild-type (WT) group. Heterozygous 5×FAD transgenic animals and WT controls were obtained after breeding progenitors purchased from the Jackson Laboratory (Jackson Laboratory, Bar Harbor, ME, USA) as described previously (Son et al., 2018).

After genotyping, the 5×FAD (TG) and age-matched WT mice were subdivided (1–2 experiments): (1) eight WT and sixteen TG mice were subjected to the behavioral tasks to characterize AD-related phenotypes of TG mice; (2) ten WT and twenty TG mice were subjected to examine memory function of TG mice after RF-EMFs; and six TG mice were used to suppress microglia.

In the first set of experiments, the TG and WT mice were then randomly assigned to three groups: (1) WT controls (WT; n = 8); (2) 5×FAD aged 6 months (TG 6M; n = 8); and (3) 5×FAD aged 8 months (TG 8M; n = 8).

To examine the effect of RF-EMFs on AD progression, the TG and WT mice were then randomly assigned to three groups: (1) WT controls (WT; n = 10); (2) 5×FAD with sham exposure (TG-C; n = 10); and (3) 5×FAD with RF-EMFs exposure (TG-RF; n = 10).

In addition, at the age of 8 months, the mice were treated with PLX3397 for 1 month to suppress microglia (TG-PLX3397; n = 6). We dissolved PLX3397 (MedKoo Biosciences, Morrisville, NC, USA; Cat# 206178) in dimethyl sulfoxide and diluted as previously described (Son et al., 2020). At 8 months old, 100 μL of the suspended PLX3397 at 50 mg/kg was administered to the TG mice daily by oral gavage for 1 month. The mice were maintained in a specific pathogen-free animal research facility and had freely access to a rodent diet and water. All animal protocols were approved by the Animal Care and Use Committee of the Korea Institute of Radiological and Medical Sciences (KIRAMS) (approval No. KIRAMS2018-0045, August 22, 2018 and KIRAMS2019-0029, May 17, 2019).

Whole-body RF-EMFs exposure system

The in vivo RF-EMFs exposure system was designed and located in a reverberation chamber (ERE-MRC-1.5; ERETEC, Gunpo, Korea). Detailed descriptions of the system, including the uniformity of the field dose, and the specific absorption rate (SAR) have been provided in previous reports (Lee et al., 2012; Son et al., 2018). Briefly, The WCDMA format code generates a 1950 MHz RF-EMFs signal from a microprocessor unit chip that controls the central processing unit. After passing through a separate digital attenuator, the signal was amplified using an additional high-power amplifier (PCS60WHPA_CW; Kortcom, Anyang, Gyeonggi-do, Korea). An 11-bit digital PIN diode attenuator (Model 349; General Microwave, Farmingdale, NY, USA) was used for controlling the output power level. A computer adjusted the exposure level and time. The external size of the reverberation chamber was 2295 mm × 2293 mm × 1470 mm; the walls were made using 2.3 mm thick stainless steel. The animals were exposed to the following schedule: SAR 5 W/kg, 2 hours/day, 5 days/week, for 6 months within the reverberation chamber installed in the animal facility. The sham exposure group was placed in the chambers for the same time except that there were no RF-EMFs signals. The mice were exposed to RF-EMFs in the reverberation chamber alternately in the morning (09:00–11:00) and afternoon (14:00–16:00) every other week to avoid a difference in their circadian rhythm. During the RF-EMFs exposure, the room temperature was maintained at 20 ± 3°C. The body temperature of the mice was measured before and immediately after RF-EMFs exposure.

Animal behavior testing

Novel object recognition test

The object recognition test was performed to assess learning and memory in mice. The procedure for the test has been described previously (Son et al., 2018). Briefly, the mice were placed on the open acrylic container (60 cm × 60 cm × 50 cm) for adaptation. The acrylic chambers and each object were cleaned with 70% (v/v) ethanol between uses to exclude the effect of odor. During the training session, we placed two objects (object A) in the arena, and the mice explored the arena for 10 minutes. Twenty-four hours after training, one familiar object (object A) was replaced by the novel object (object B) for testing. The camera was placed above the arena, and the movements of the mice were recorded for video tracking (SMART 3.0; Panlab, Barcelona, Spain). Each preference percentage was defined as the number of visits with a specific object divided by the total number of visits with either object.

Y-maze test

Detailed procedures for Y-maze test have been described previously (Son et al., 2016). The Y-maze test is widely used to test the working and reference memory based on the recording of spontaneous alternation behavior. Each mouse was placed into the center of the Y-maze, and the movement of each mouse was recorded by a computer program (SMART 3.0) for 8 minutes. Alternation behavior was analyzed as entries into the three arms on overlapping triplet sets and scored as the ratio of actual to possible alternations (defined as the total number of arm entries minus two) multiplied by 100, as follows: % alternation = number of alternations/(total arm entries – 2) × 100.

Western blotting

After behavioral tests, mice were euthanized by CO₂ asphyxia with 30–70% of the flow introduction rate (chamber volume/minute) and brain hippocampal tissues were collected and immediately stored at –70°C. Hippocampi were sonicated in lysis buffer (PRO-PREP™, iNtRon, Gyeonggi-do, Korea) by physical mincing. Sample buffer was added after protein quantification, and the samples were denatured by heating at 100°C for 5 minutes in boiling water. The proteins of each sample were separated by electrophoresis on 10–15% SDS-acrylamide gels using 1:29 bis/acrylamide (MBiotech, Gyeonggi-do, Korea, Cat# 22100) and then transferred onto nitrocellulose membranes (Pall Corporation, Port Washington, NY, USA, Cat# 66485). For non-specific binding, the membranes were blocked with a 5% bovine serum albumin (Gendepot, Katy, TX, USA, Cat# A0100-010). The membranes were then incubated with primary antibodies, including mouse anti-6E10 (1:1000, Covance, Princeton, NJ, USA, Cat# Sig-39320, RRID: AB_662798, revealing full-length APP, C99, and Aβ), rabbit anti-alpha-secretase (ADAM10; 1:1000, Abcam, Cambridge, UK, Cat# ab1997, RRID: AB_302747), mouse anti-beta-secretase 1 (BACE1; 1:1000, Santa Cruz Biotechnology, Dallas, TX, USA, Cat# sc-33711, RRID: AB_626716), goat anti-glial fibrillary acidic protein (GFAP; 1:1000, Abcam, Cat# ab53554, RRID: AB_880202), mouse anti-Iba1 (1:1000, Abcam, Cat# ab15690, RRID: AB_2224403), rabbit anti-CSF1R (1:1000, Santa Cruz Biotechnology, Cat# sc-692, RRID: AB_631025), and mouse anti-β-actin (1:3000, Cell Signaling Technology, Danvers, MA, USA, Cat# 3700, RRID: AB_2242334), overnight at 4°C. After extensive washing in PBS-T buffer, the membranes were incubated for 1 hour at room temperature with the corresponding horseradish peroxidase (HRP)-conjugated secondary antibody; goat anti-rabbit IgG-HRP (1:10000, Abcam, Cat# ab97051, RRID: AB_10679369), goat anti-mouse IgG-HRP (1:10000, Abcam, Cat# ab205719, RRID: AB_2755049), donkey anti-goat IgG-HRP (1:3000, Santa Cruz Biotechnology, Cat# sc-2020, RRID: AB_631728). Protein bands were visualized using a chemiluminescence kit (Perkin Elmer, Waltham MA, USA) and then quantified using ImageJ software (version 1.51, National Institutes of Health (NIH), Bethesda, MD, USA; Schneider et al., 2012). Relative protein expression was normalized to β-actin.

Immunofluorescence staining and quantification

The whole brain was fixed in 4% paraformaldehyde and then embedded in paraffin block. The brain sample from each mouse was sectioned at approximately 2 mm posterior to the bregma, and a standardized counting area that contained 4-μm-thick coronal sections cut by a rotary microtome (HM325, Thermo Fisher Scientific, Waltham, MA, USA) was used. Two non-overlapping sections of the two regions of the hippocampus, approximately 50 μm apart, were analyzed. After deparaffinization and hydration, each section was incubated with citrate buffer (H-3300-250, Vector Laboratories Inc., Burlingame, CA, USA) for heat-induced antigen retrieval. Thereafter, the brain sections were incubated in normal goat serum to block nonspecific reactions and reacted with primary antibodies mouse anti-6E10 (1:1000, Covance, Cat# Sig-39320, RRID: AB_662798), goat anti-Iba1 (1:200, Abcam, Cat# ab48004, RRID: AB_2755049), and rabbit anti-CSF1R (1:100, Santa Cruz Biotechnology, Cat# sc-692, RRID: AB_631025) overnight at 4°C. Alexa Fluor™ 488-conjugated donkey anti-rabbit IgG (1:200, Invitrogen, Waltham, MA, USA, Cat# A-21206, RRID: AB_2535792) and Alexa Fluor™ 546-conjugated donkey anti-mouse IgG (1:200, Invitrogen, Cat# A10036, RRID: AB_2534012) or Alexa Fluor™ 488-conjugated donkey anti-goat IgG (1:200, Cat# A-11055, Invitrogen, RRID: AB_2534102) and Alexa Fluor™ 546-conjugated donkey anti-rabbit IgG, (1:200, Cat# A10040, Invitrogen, RRID: AB_2534016). Following incubation, the sections were counterstained with 4′,6-diamidino-2-phenylindole (Thermo Fisher Scientific, Cat# D1306, RRID: AB_2629482) and coverslipped. The stained sections were observed and imaged using a fluorescence microscope (BX-53, Olympus, Tokyo, Japan).

Immunofluorescence images were captured using a BX-53 microscope (Olympus, Tokyo, Japan) equipped with a CCD DP73 camera. For quantification, the images were converted to grayscale, and the threshold of each image was adjusted via background subtraction. The mean gray value (256 Gy levels) for each selected area was determined using ImageJ software (version 1.51). NIH ImageJ particle counting plug-in was used to calculate the number of Aβ burden (plaques) per unit area. Only plaques > 10 µm in diameter were analyzed to avoid the possibility of noise/debris being included. All of the measurements were performed by the same individual, who was blinded to the experimental conditions.

Quantitative reverse transcription-polymerase chain reaction

For RNA extraction QIAzol Lysis Reagent (Qiagen, Valencia, CA, USA, Cat# 79306) was used, and cDNA was synthesized by using cDNA Synthesis Platinum Master Kit (Gendepot, Cat# R5600), as described by the manufacturer. Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was carried out using SYBR Green 2X Mastermix kit (MBiotech, Gyeonggi-do, Korea, Cat# 18303) and gene-specific primers. The thermal cycling profile consisted of a preincubation step at 95°C for 10 minutes, followed by 45 cycles of denaturation (95°C, 30 seconds) and annealing and elongation (60°C, 30 seconds). For housekeeping gene in relative gene expression, glyceraldehyde 3-phosphate dehydrogenase (Gapdh) was used. Results are normalized to Gapdh gene expression using the 2^–ΔΔCt method. The sequences for each primer are listed in Table 1. For the qRT-PCR experiments, the CFX96 real-time PCR detection system (Bio-Rad, Hercules, CA, USA) was used.

Table 1.

Primer used in quantitative reverse transcription-polymerase chain reaction analysis

Gene	Forward sequence (5’–3’)	Reverse sequence (5’–3’)	Accession number
Ccl6	AGA TCG TCG CTA TAA CCC TC	ATC TGT GTG GCA TAA GAG AAG	NM_009139.3
CD14	CTT AGT CAC AAT TCA CTG CG	AAG ACA GAT TGA GCG AGT TT	NM_009841.4
CD68	CCA CAG GCA GCA CAG TGG ACA	TCC ACA GCA GAA GCT TTG GCC C	NM_001291058.1
Clu	CAT AGA CAC GCT CTT TCA GG	GAT GGG GGA GAA GTA GTG G	NM_013492.3
Csf1r	CCC TAC TCA GTT GCC CTA CA	GCC TCC TTC TCA TCA GCA TG	NM_001037859.2
Ctsb	ACC GTA CCT GTA ACT GCT CAC	CAC TCC AGG GAA ATG ACC TCT AA	NM_007798.3
Ctsd	CTG AAG CAG GCA AAT GGG TC	CAG AGT TGG AGG GGC AGT AG	NM_009983.3
Ctss	GAG CAC CAC ACT TCA GGA TGA	CAA TGG TAG TCC AGG GTA GGG	NM_001267695.2
Fcgr1	GAG TAC CAT ATA GCA AGG GC	AAA CAG GAT GTG AAA CCA GA	NM_010186.5
Gapdh	CAA GAA GGT GGT GAA GCA GG	AGG TGG AAG AGT GGG AGT TG	NM_008084.3
IL-1β	ACC TTT TGA CAG TGA TGA GAA	GCT GCT GCG AGA TTT GA	NM_008361.4
IL-4	AAG AAC ACC ACA GAG AGT GA	ATG AAT CCA GGC ATC GAA AA	NM_021283.2
IL-6	CCT TCC CTA CTT CAC AAG TC	TTT TCT GCA AGT GCA TCA TC	NM_001314054.1
Mmp2	AGC TGT ACA GAC ACT GGT CG	GTA AAC AAG GCT TCA TGG GGG	NM_008610.3
Mmp9	CAA AGG CAG CGT TAG CCA G	GGA AGA CCA CAA AAG TCG GC	NM_013599.4
Spi1	GGA TTT CTC CGC ACA CCA TG	GCA TCT GTT CCA GCT CCA TG	NM_011355.2
Timp1	CCC CAG AAA TCA ACG AGA CCA	TAC CGG ATA TCT GCG GCA TT	NM_001044384.1
Timp2	AGG AGA TGG CAA GAT GCA CA	GAT CAT GGG ACA GCG AGT GA	NM_011594.3
Trem2	AGG TTT CAT CCT GTG GGT CA	AGG AGG TCT CTT GAT TCC TTG	NM_001272078.1

Clu: Clusterin; Csf1r: colony-stimulating factor 1 receptor; Ctsb: cathepsin B; Ctsd: cathepsin D; Gapdh: glyceraldehyde 3-phosphate dehydrogenase; IL: interleukin.

Statistical analyses

No statistical methods were used to predetermine sample sizes; however, our sample sizes are similar to those reported in previous publications (Zhang et al., 2017; Qubty et al., 2021). The data are represented as the mean ± standard error of the mean. The data were analyzed by one-way analysis of variance followed by Tukey’s post hoc test using GraphPad Prism 8.0 (GraphPad Software, San Diego, CA, USA). For all analyses, P < 0.05 was determined to indicate statistical significance.

Results

Progressive memory impairment and neuroinflammation in a mouse model of 5×FAD

5×FAD, a transgenic AD mouse model, exhibits many AD-related phenotypes and a relatively progressive and aggressive Aβ production. Immunofluorescent staining for Aβ showed that the number of Aβ plaques increased in TG mice with aging, as expected (Figure 1A and B). Immunostaining for the microglia marker, Iba1, showed increases in microglial densities and activated morphology at 6 and 8 months of age in the hippocampus of 5×FAD mice compared with WT mice at the age of 8 months. Activated microglia were more densely clustered around core plaque with aging in 5×FAD mice.

Figure 1.

Progressive amyloid-β pathology and cognitive dysfunctions in 5×FAD mice.

(A) Representative immunofluorescence images of microglia (Iba1 in green) and amyloid-β (Aβ; 6E10 in red) in the hippocampi of wild-type (WT) or 5×FAD (TG) mice. Yellow arrowheads indicate activated microglia. (B) Quantification of the expression levels of amyloid-beta burden in the hippocampi of mice. (C) In the novel object recognition test, the TG mice showed a significantly reduced percentage of preference for novel objects (object B) than the familiar objects (object A) in the testing session. (D) In the Y-maze test, the TG mice showed fewer spontaneous alternations than the WT controls. (E) The graph shows the percentages of IL-1β and IL-6 expression in TG and WT mice as examined by quantitative reverse transcription-polymerase chain reaction. The data are shown as the mean ± standard error of the mean. *P < 0.05, ***P < 0.001 (n = 8). M: Month; SAP: spontaneous alternation performance; TG: 5×FAD; WT: wild-type.

To examine cognitive functions in TG and WT mice, we performed the novel object recognition memory and Y-maze tests in WT and TG mice at the age of 6 and 8 months. At the age of 8 months old, the TG mice showed more severe cognitive impairments in preferential recognition of the novel object (Figure 1C) and reduced spontaneous alternation in the Y-maze test as well when compared with those of 6 months old, indicating that 5×FAD mice showed age-dependent behavioral deficits (Figure 1D). In addition, we analyzed the mRNA expression of genes encoding proinflammatory cytokines, including IL-1β and IL-6, in the mouse hippocampus. Compared to those in WT mice, the expression levels of IL-1β and IL-6 were significantly increased in the TG mouse hippocampus (Figure 1E; P < 0.05). These observations showed that cognitive dysfunctions in 5×FAD mice were progressive and worsen with age, which was accompanied by increases in levels of inflammation-related genes in the hippocampus.

Effects of RF-EMFs exposure on memory function in a mouse model of AD

To evaluate the effect of RF-EMFs exposure on behavioral changes, mice were exposed to RF-EMFs at an SAR of 5 W/kg for 2 hours/day and 5 days/week for 6 months (Figure 2A). The novel object recognition memory and Y-maze tests were performed to examine cognitive impairment in TG and WT mice at the age of 8 months old after the termination of RF-EMFs exposure. During the training session of the object recognition memory test, the WT, TG, and TG-RF groups displayed equal preferences for the two objects (Figures 2B and C). Twenty-four hours after the training session, mice of the TG group showed less of a preference for the novel object than the WT mice; however, the cognitive impairments in TG mice were mitigated by RF-EMFs exposure, as reflected by the preference increase to the levels observed in WT mice.

Figure 2.

Effects of radiofrequency electromagnetic field exposure on cognitive function in 5×FAD mice.

(A) Schematic diagram of our experimental procedure. (B) Representative movement trajectory images of mice in the novel object recognition memory test. (C) The percentages of visits during the training (upper) and testing (lower) sessions were evaluated. (D) The representative diagrams for the spontaneous alternation in the Y-maze test. (E) The alternation percentages of each group in the Y-maze test. The data are shown as the mean ± standard error of the mean. *P < 0.05 (n = 10). SAP: Spontaneous alternation performance; TG-C: sham-exposed TG; TG-RF: radiofrequency electromagnetic fields-exposed TG; WT: wild-type.

To evaluate whether RF-EMFs exposure affects short-term working memory, we subjected mice exposed to RF-EMFs for 6 months to the Y-maze (Figure 2D). Compared with the WT controls, the TG group mice showed a significant decrease in spontaneous alternations in the Y-maze test (Figure 2E). The decrease in spontaneous alternation was rescued in the TG-RF group compared with the TG group (Figure 2E). These results indicate that RF-EMFs exposure, as examined by the novel object recognition memory and Y-maze tests, might contribute to the improvement of memory function in a mouse model of AD.

Effects of RF-EMFs exposure on APP processing in an AD mouse model

To demonstrate the effects of RF-EMFs on Aβ accumulation, we first examined the levels of Aβ deposition by using 6E10 staining. Immunofluorescence analysis revealed decreases in Aβ plaque deposition in the hippocampi of the TG-RF group compared with the TG group (Figure 3A). To further examine the effects of RF-EMFs on APP processing, Western blotting was performed to analyze the protein expression in the hippocampi of 5×FAD mice (Figure 3B). The APP, C99, and Aβ levels were increased significantly in TG mice compared with the age-matched WT mice. Although the APP and C99 protein levels were not significantly altered by RF-EMFs exposure, the Aβ expression in 5×FAD mice exposed to RF-EMFs was significantly decreased compared with that in TG mice (Figure 3C). Apart from that, the BACE1 and ADAM10 levels did not differ between TG mice and those of the WT and TG-RF groups (Figure 3C). These results suggest that RF-EMFs exposure is sufficient to decrease the production and deposition of Aβ.

Figure 3.

Effects of RF-EMFs on the levels of Aβ and APP processing in the hippocampi of mice.

(A) Representative immunofluorescence images and quantification of Aβ (6E10 in red) in the hippocampi of 5×FAD mice after RF-EMFs exposure. (B and C) Representative images of Western blotting and the expression levels of APP, C99, Aβ, beta-secretase 1 (BACE1), and alpha-secretase (ADAM10) in the hippocampi of mice in the WT, TG-C, and TG-RF groups. The values are shown as the mean ± standard error of the mean. *P < 0.05, **P < 0.01 (n = 4 for A, n = 6 for C). Aβ: Amyloid-beta; APP: amyloid precursor protein; ns: not significant; TG-C: sham-exposed TG; TG-RF: radiofrequency electromagnetic fields-exposed TG; WT: wild-type.

Changes in the levels of Aβ-related genes following RF-EMFs exposure

To confirm the effects of RF-EMFs exposure on Aβ, we analyzed the mRNA expression of genes specifically related to Aβ in the hippocampal region of the mouse brain. The mRNA expression levels of cathepsin B (Ctsb) and cathepsin D (Ctsd) were not altered by either transgenic expression or RF-EMFs exposure as determined by quantitative PCR (Figure 4A). The expression levels of CD14 and clusterin (Clu) were significantly increased in TG mice, but their levels did not differ between the TG-RF and TG groups (Figure 4B). The expression levels of Aβ-degrading enzymes, including Mmp2 and Mmp9, were not altered by RF-EMFs exposure (Figure 4C and D). Similarly, no differences in other genes associated with Aβ clearance (Timp1 and Timp2) were observed following RF-EMFs (Figure 4C and D). These data suggest that RF-EMFs exposure does not change the expression levels of genes associated with Aβ processing.

Figure 4.

Changes in the expression levels of amyloid beta-related genes in the hippocampus following radiofrequency electromagnetic fields exposure.

The data show the levels of mRNAs encoding cathepsin B (Ctsb) and cathepsin D (Ctsd) (A) as well as CD14 and clusterin (Clu) (B) in the hippocampi of 5×FAD mice. The mRNA levels of matrix metalloproteinase 2 (Mmp2), tissue inhibitor of metalloproteinase 2 (Timp2) (C) as well as those of Mmp9 and Timp1 (D) were evaluated in the mice. The values are shown as the mean ± standard error of the mean. **P < 0.01, ***P < 0.001 (n = 6). Aβ: Amyloid-beta; APP: amyloid precursor protein; TG-C: sham-exposed TG; TG-RF: radiofrequency electromagnetic fields-exposed TG; WT: wild-type.

Effects of RF-EMFs exposure on microglial and astrocytic activation

To determine whether RF-EMFs exposure prevents the activation of microglia, we examined the levels of Iba1 (a microglial marker) and CSF1R (a microglia proliferative marker) by using specific markers. The number of Iba1-positive and Iba1/CSF1R-double positive cells in the hippocampi of each group were counted (Figure 5A and B). In the TG group, there are significant increases in Iba1- and Iba1/CSF1R-expressed microglia compared to the WT group. As known that CSF1R is expressed in microglia (Hagan et al., 2020; Han et al., 2022), CSF1R expression is overlapped in most of Iba1-positive microglia as seen in the inserted image of Figure 5A. The number of microglia expressing CSF1R remarkably reduced in the hippocampus of the TG-RF group. The expression levels of Iba1 and GFAP, a biomarker of astrocytes, were increased in the hippocampal region of TG mice compared with WT mice as determined by Western blotting (Figure 5C and D); RF-EMFs significantly reduced the expression of Iba1 in the TG-RF group compared with the sham-exposed TG group. However, the difference in the GFAP levels did not reach statistical significance (Figure 5C and D).

Figure 5.

Effects of radiofrequency electromagnetic fields on Iba1 and CSF1R in the hippocampi of 5×FAD mice.

(A) Representative images of microglia (Iba1 in green) and CSF1R (red) in the hippocampi of 5×FAD mice after radiofrequency electromagnetic fields exposure. The yellow square shows the merged images at higher magnification (Inserted panel). (B) The quantification of the Iba1- and Iba1/CSF1R-positive cells in the mice hippocampus. (C and D) Representative images for Western blots and the expression levels of Iba1 and GFAP in the hippocampi of mice in the WT, TG, and TG-RF groups. (E and F) Representative images for Western blots and the expression levels of CSF1R in the hippocampi of mice in the WT, TG, and TG-RF groups. The values are shown as the mean ± standard error of the mean. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (n = 4 for B, n = 6 for D and F). CSF1R: Colony-stimulating factor 1 receptor; GFAP: glial fibrillary acidic protein; Iba1: ionized calcium-binding adapter molecule 1; ns: not significant; TG-C: sham-exposed TG; TG-RF: radiofrequency electromagnetic fields-exposed TG; WT: wild-type.

Next, we investigated whether the expression of CSF1R was altered by RF-EMFs exposure. Western blot analysis revealed significant increases in the levels of CSF1R in the TG mouse hippocampus compared with WT mice (P < 0.001; Figure 5E and F). However, these increases were significantly ameliorated in the TG-RF group relative to the TG group (P < 0.05; Figure 5E and F), which was consistent with the expression of Iba1, suggesting that neuroinflammation was ameliorated by RF-EMFs exposure in the AD mouse model.

Transcriptional changes involved in microgliosis and inflammatory cytokines by RF-EMFs exposure compared to PLX3397, a CSF1R inhibitor

As neuroinflammation is an important feature of AD, we next examined whether the changes in the levels of inflammatory signaling molecules were related to microglial depletion. Compared with those in WT mice, the expression levels of Csf1r, CD68, and Ccl6 were significantly increased in the hippocampi of TG mice (Figure 6A). However, the levels of these genes were significantly down-regulated in the hippocampi of 5×FAD mice after both RF-EMFs and PLX3397 treatment (Figure 6A). The expression levels of genes encoding proinflammatory cytokines, including IL-1β and IL-6, were up-regulated significantly in TG mice (Figure 6B). However, IL-1β gene expression was almost erased by RF-EMFs exposure; this effect was even observed after PLX3397 treatment. In addition, RF-EMFs exposure downregulated the expression of IL-6 in both the TG-RF and TG-PLX3397 groups compared with the vehicle-treated TG group, although statistical significance was not reached in the TG-RF group (Figure 6B). The mRNA levels of IL-4 were not altered by either transgenic expression or RF-EMFs exposure (Figure 6B). These results suggested that RF-EMFs exposure reduced the levels of inflammatory-related genes in the hippocampi of 5×FAD mice, which was consistent with Csf1r being inhibited in 5×FAD mice by PLX3397 treatment.

Figure 6.

Changes in the expression levels of genes related to microgliosis and inflammatory cytokines in the hippocampus following radiofrequency electromagnetic fields exposure.

The data show the levels of mRNAs encoding (A) microgliosis (colony-stimulating factor 1 receptor [Csf1r], CD68, and Ccl6) and (B) inflammatory cytokines (interleukin [IL]-1β, IL-6, and IL-4) in the hippocampi of 5×FAD mice. The values are shown as the mean ± standard error of the mean. *P < 0.05, **P < 0.01, ***P < 0.001 (n = 6). ns: Not significant; TG-C: sham-exposed TG; TG-RF: radiofrequency electromagnetic fields-exposed TG; TG-PLX3397: PLX3397-treated TG; WT: wild-type.

Changes in the levels of transcripts associated with microglia-related genes after RF-EMFs exposure

To determine the impact of RF-EMFs exposure, the expression levels of microglia-related genes (Trem2, Spi1, Ctss, and Fcgr1) were evaluated in the mouse brain hippocampus. Compared with the WT control, the TG group exhibited significantly increased expression levels of Trem2, Fcgr1, Ctss, and Spi1 in the hippocampus after 6 months of RF-EMFs exposure (Figure 7A–D). However, the expression levels of genes associated with the microglial function (Trem2 and Fcgr1) were markedly decreased in the TG-RF group, consistent with the TG-PLX3397 group (Figure 7A and B). In addition, the Ctss and Spi1 expression levels were decreased in both the TG-RF and TG-PLX3397 groups compared to the age-matched TG mice (Figure 7C and D). These findings reveal marked decreases in the mRNA expression levels of genes related to microglia following RF-EMFs.

Figure 7.

Changes in the mRNA expression levels of genes related to microglial function in the hippocampus following radiofrequency electromagnetic fields exposure.

The bar graphs show the levels of mRNAs encoding Trem2 (A), Fcgr1 (B), Ctss (C), and Spi1 (D) in the hippocampi of 5×FAD mice. The values are shown as the mean ± standard error of the mean. *P < 0.05, **P < 0.01, ***P < 0.001 (n = 6). TG-C: Sham-exposed TG; TG-RF: radiofrequency electromagnetic fields-exposed TG; TG-PLX3397: PLX3397-treated TG; WT: wild-type.

Discussion

This study demonstrated that 6M RF-EMFs exposure led to decreases in Aβ levels in a mouse model of AD. Although RF-EMFs exposure did not change the levels of genes related to Aβ degradation, it downregulated the levels of microglial markers in the mouse hippocampus. In addition, altered expression of the inflammatory-related genes was found in AD mice following RF-EMFs exposure, which was consistent with the changes in genes related to microglial function. Therefore, the current study strongly provides a possible new mechanism regarding RF-EMFs action that may contribute substantially to it is potential to ameliorate or reverse AD.

Because no effective AD therapeutics have been found to slow or reverse the behavioral dysfunctions induced by AD thus far, nonpharmaceutical applications are now being considered as a possible alternative for AD treatment. Many research groups are investigating the effects of RF-EMFs on cognitive behavior in mice. The results are controversial as there was primarily no alteration and a few adverse or useful effects on cognitive behaviors in non-disease states (Sienkiewicz and van Rongen, 2019). Interestingly, it has been reported that most of the therapeutic effects of RF-EMFs include neuropathological and behavioral alterations in AD models, including 3×Tg-AD mice (Banaceur et al., 2013), 5×FAD mice (Jeong et al., 2015; Son et al., 2018), and APPsw mice (Arendash et al., 2010; Dragicevic et al., 2011; Mori and Arendash, 2011). Some scholars have reported that EMF treatment (pulsed at 918 MHz, 1.05 W/kg SAR) provided a beneficial effect on the cognitive dysfunction of AD mice (APPsw Tg or APP + PS1 Tg mice) (Mori and Arendash, 2011; Arendash et al., 2012); they proposed the mechanisms of EMF action as anti-Aβ aggregation, mitochondrial enhancement and increased neuronal activity in the brain (Arendash, 2012). Perez et al. (2021) also reported that repeated electromagnetic field stimulation (64 MHz, 1 hour/day) at SAR > 0.4 W/kg significantly reduced Aβ levels without any toxic effects in primary human brain cultures. In addition, in clinical trials, cognitive enhancement and changes in blood AD marker levels were found in AD patients subjected to electromagnetic treatment for 2 months (Arendash et al., 2019). Previously, we have reported the beneficial effects of RF-EMFs exposure (1950 MHz, SAR 5 W/kg, 2 hours/day, 5 days/week, for 8 months) in AD transgenic mice (5×FAD mice) by analyzing cognitive functions and the Aβ deposition in the entorhinal cortex and hippocampus (Jeong et al., 2015) and glucose uptake in the hippocampus (Son et al., 2018). This indicates that RF-EMFs may be useful for the brain in diseased states. In addition, since most patients with dementia are old and diagnosed with the disease after the cognitive decline, therapeutic modalities would begin in patients with progressive lesions, such as Aβ plaques and neuroinflammation. Therefore, RF-EMFs could be applied as a treatment or palliative modality for progressive disease.

In the present study, 6 months of RF-EMFs exposure improved the performances of mice on the object recognition memory and Y-maze tests, which was consistent with the decreased Aβ expression in the hippocampus. These results demonstrated that RF-EMFs exposure ameliorates cognitive impairment in AD mice. In the brains of AD patients and murine models, Aβ and Aβ-related genes are one of the pathways correlated with AD progression. A recent study demonstrated that Ctsb and Ctsd are strongly implicated in the pathogenesis of AD and related to the degradation of Aβ (Oberstein et al., 2020; Suire et al., 2020). Moreover, the levels of the CD14 and Clu genes were reported to be increased in AD patients and associated with Aβ aggregation and clearance (Foster et al., 2019; Pase et al., 2020). A previous study demonstrated that transgenic mice lacking CD14 showed increased levels of inflammatory cytokines, indicating that CD14 might modulate the inflammatory processes in the AD brain (Reed-Geaghan et al., 2010). In addition, Clu interacts with Aβ and plays important roles in the clearance of soluble Aβ from the brain (Wojtas et al., 2017). In the present study, the expression of CD14 and Clu was increased in the AD model (TG) compared with the WT control, which was consistent with a previous study (Spangenberg et al., 2019). However, altered patterns of Ctsb, Ctsd, CD14, and Clu expression were not observed after RF-EMFs exposure. Previously, Mmp2 and Mmp9 were shown to be related to regulating Aβ cleavage, and many investigations have focused on modulating these genes in neurodegenerative and neuroinflammatory diseases (Agrawal et al., 2006). In the present study, no alterations in Mmp2 and Mmp9 or its negative regulators Timp1 and Timp2 were observed in the AD mice after RF-EMFs exposure. Collectively, these data indicate that the alteration of Aβ expression might not be related to the changes in the genes related to Aβ cleavage in the AD mouse model.

It is generally accepted that AD is associated with robust microgliosis and that neuroinflammation might critically contribute to AD pathogenesis. In neurodegenerative conditions such as AD, microglial cells are increased and become dysfunctional upon the impairment of phagocytotic processes (Mosher and Wyss-Coray, 2014). In the present study, although the protein expression of Iba1 and GFAP was increased in the AD mice, which was consistent with the previous study (Jeong et al., 2015), the increased expression of Iba1 in AD mice was alleviated following RF-EMFs exposure. This phenomenon might be related to changes in the state of microglia and the spectrum of produced cytokines. To investigate the hypothesis regarding the reduction in microgliosis after RF-EMFs exposure, we performed qRT-PCR analysis of 5×FAD mice in which microglia were suppressed by PLX3397. Previously, CSF1R was shown to play an important role in microglial proliferation during the progression of chronic neurodegenerative diseases, including AD (Gomez-Nicola et al., 2013). In addition, previous studies have reported the effect of PLX3397, a CSF1R inhibitor, on 5×FAD mice of various ages (Sosna et al., 2018; Son et al., 2020). In the present study, the qRT-PCR results demonstrated that RF-EMFs exposure prevented the expression of Csf1r, with a decrease in the levels of CD68. These phenomena were also observed in 5×FAD mice treated with PLX3397. In addition, RF-EMFs exposure downregulated the levels of proinflammatory cytokines, including IL-1β and IL-6, and upregulated those of the anti-inflammatory cytokine IL-4, within the hippocampus, which might indirectly indicate the suppression of microglial activity.

There have been reports on the beneficial effect of RF-EMFs on cognitive behaviors and possible mechanism-related Aβ proteins, neuronal activity, and mitochondrial functional enhancement in the AD brain by a few research groups (Sienkiewicz and van Rongen, 2019). To corroborate the notion that RF-EMFs as a new strategy for AD treatment, it should be supported by clear evidence representing the direct mechanism and effectiveness of the therapeutic effects of RF-EMFs on AD. To date, there have been no reports investigating the neuroinflammatory response to RF-EMFs exposure. Our study suggests the potential impact of RF-EMFs on the regulation of microglial dysfunction and its anti-inflammatory effects on microgliosis in the AD brain. In many neurodegenerative diseases, microglial activation is a hallmark of lesions, accompanied by an increased number of microglia and dysfunctional phenotypes of microglia (Heneka et al., 2014). Our results suggest that RF-EMFs can prevent microgliosis caused by Aβ pathology. RF-EMFs reduced not only the total number of activated microglia represented by upregulated Iba1 and CSF1R, but also transcription-related dysfunctional phenotypes of microglia, such as CSF1R or TREM2. Our study did not show the direct regulation of microglia-modifying factors, such as CSF1R, by RF-EMFs, although RF-EMFs showed potent inhibitory effects on microgliosis in AD. Therefore, further studies are needed to investigate the direct mechanisms regulating microglial dysfunction and gliosis. As the regulation of microglia is a potential target for the prevention or treatment of neurodegenerative diseases that currently have no effective clinical treatment, RF-EMFs could be an alternative therapeutic modality for these diseases.

To demonstrate potential associations between microglial function and cognitive dysfunction in AD mice, we focused on the impact of RF-EMFs on the expression of genes associated with microglial function as determined by qRT-PCR analysis. Microglia play a crucial role in the homeostasis conditions of the brain, and understanding microglial functions is important for elucidating strategies for the treatment of neurodegenerative diseases. A previous study investigated the potential involvement of several genes, including Trem2, Fcgr1, Ctss, and Spi1, in microglial function and their expression in AD brain tissues (Rustenhoven et al., 2018). Ctss was shown to play a role in the pathogenesis of neurodegenerative diseases, including AD (Schechter and Ziv, 2011), and Ctss inhibitors have been utilized as a treatment strategy to prevent neurodegenerative disorders (Haque et al., 2008). Moreover, Spi1 is reportedly expressed in brain microglia, and attenuation of Spi1 regulates AD by reducing neuroinflammatory responses (Rustenhoven et al., 2018). We found that the levels of Trem2, Fcgr1, Ctss, and Spi1 were increased significantly in AD mice and that these increases were ameliorated by RF-EMFs exposure, which was correlated with the downregulation of gene expression after PLX3397. Collectively, these results suggested that the upregulated expression of genes related to microglial function in the AD mouse model was effectively ablated by both RF-EMFs and PLX3397, indicating that the beneficial effects of RF-EMFs on the cognitive functions of AD mice might be related to the inhibition of microglial functions (Figure 8).

Figure 8.

Schematic illustration of therapeutic effects of RF-EMFs exposure on the brain of AD mice.

(A) AD-associated morphological changes of microglia and microglial-related genes. (B) Proposed mechanism of RF-EMFs action on AD. In the brain of AD mice, microglial dysfunction and neuroinflammation are induced, which result in disturbances of memory processing. In activated microglia of AD mice, the expression levels of microglial-related genes (Trem2, Fcgr1, Spi1, and Ctss) and pro-inflammatory genes are up-regulated. RF-EMFs exposure has the potential to positively restore cognitive impairments in AD mice via the modulation of microglial-related and neuroinflammation-related genes. AD: Alzheimer’s disease; IL: interleukin; RF-EMFs: radiofrequency electromagnetic fields.

A 3-month exposure to RF-EMFs did not have a significant preventive effect on the impaired memory behaviors and Aβ pathology of 5×FAD (Son et al., 2016), while eight-month RF-EMFs exposure showed remarkable preventive effects on hippocampal-dependent memory impairment and Aβ pathology, including Aβ deposition, gliosis, and reduced glucose metabolism (Jeong et al., 2015; Son et al., 2018). In the current study, the six-month exposure of 5×FAD mice to RF-EMFs was revealed to be effective in inhibiting Aβ load and neuroinflammation, including microgliosis. Therefore, we speculate that long-term treatment would be effective for AD, and the duration could be longer than a few months depending on disease progression and SAR levels. Alternatively, it can be considered as a treatment for the head with a high intensity of RF-EMFs (higher SAR). Since RF-EMFs can deliver energy to the body, RF-EMFs with high SAR may induce thermal effects on the body. To avoid the whole-body thermal effect, we applied SAR 5 W/kg, which did not cause an increase in body temperature. Further studies with high-intensity (SAR) and local application to the head region are needed for the therapeutic application of RF-EMFs in AD.

Together, the results of our study demonstrated the effects of RF-EMFs exposure on the memory performance and Aβ burden of AD model mice. We found that RF-EMFs exposure preserved hippocampal-dependent memory function in an AD mouse model with remarkable suppression of neuroinflammation. Consequently, we suggest that the therapeutic effect of RF-EMFs on hippocampal dysfunction in AD model mice is related to the decreased levels of neuroinflammation- and microglial function-related signals, concomitant with a microglial reduction in the hippocampus. These results indicate that RF-EMFs may have a therapeutic effect on AD via suppressing activated microglia and suggest that additional studies be performed to further develop neurodegenerative disease management strategies.