Glyphosate exposure exacerbates neuroinflammation and Alzheimer’s disease-like pathology despite a 6-month recovery period in mice

Abstract

Background

Glyphosate use in the United States (US) has increased each year since the introduction of glyphosate-tolerant crops in 1996, yet little is known about its effects on the brain. We recently found that C57BL/6J mice dosed with glyphosate for 14 days showed glyphosate and its major metabolite aminomethylphosphonic acid present in brain tissue, with corresponding increases in pro-inflammatory cytokine tumor necrosis factor-⍺ (TNF-⍺) in the brain and peripheral blood plasma. Since TNF-⍺ is elevated in neurodegenerative disorders such as Alzheimer’s Disease (AD), in this study, we asked whether glyphosate exposure serves as an accelerant of AD pathogenesis. Additionally, whether glyphosate and aminomethylphosphonic acid remain in the brain after a recovery period has yet to be examined.

Methods

We hypothesized that glyphosate exposure would induce neuroinflammation in control mice, while exacerbating neuroinflammation in AD mice, causing elevated Amyloid-β and tau pathology and worsening spatial cognition after recovery. We dosed 4.5-month-old 3xTg-AD and non-transgenic (NonTg) control mice with either 0, 50 or 500 mg/kg of glyphosate daily for 13 weeks followed by a 6-month recovery period.

Results

We found that aminomethylphosphonic acid was detectable in the brains of 3xTg-AD and NonTg glyphosate-dosed mice despite the 6-month recovery. Glyphosate-dosed 3xTg-AD mice showed reduced survival, increased thigmotaxia in the Morris water maze, significant increases in the beta secretase enzyme (BACE-1) of amyloidogenic processing, amyloid-β (Aβ) 42 insoluble fractions, Aβ 42 plaque load and plaque size, and phosphorylated tau (pTau) at epitopes Threonine 181, Serine 396, and AT8 (Serine 202, Threonine 205). Notably, we found increased pro- and anti-inflammatory cytokines and chemokines persisting in both 3xTg-AD and NonTg brain tissue and in 3xTg-AD peripheral blood plasma.

Conclusion

Taken together, our results are the first to demonstrate that despite an extended recovery period, exposure to glyphosate elicits long-lasting pathological consequences. As glyphosate use continues to rise, more research is needed to elucidate the impact of this herbicide and its metabolites on the human brain, and their potential to contribute to dysfunctions observed in neurodegenerative diseases.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12974-024-03290-6.

Introduction

The increased prevalence of neurodegenerative disorders such as Alzheimer’s Disease (AD) is alarming given that more than 6.7 million people are affected with this disorder in the US alone, and this number may reach 14 million Americans by 2060 [1]. People with AD present with severe memory loss, diminished decision-making abilities, and other behavioral changes that affect their everyday life [1]. The key pathologies present in AD that subsequently disrupt neuronal function leading to cognitive deficits include the accumulation of (1) extracellular amyloid beta (Aβ) plaques, (2) intracellular neurofibrillary tau tangles (NFT), and (3) neuroinflammation [1–3]. An abundance of work highlights that environmental factors play a role in sporadic AD, which accounts for > 95% of those affected with AD overall [4, 5]. Factors such as air pollution, diet, exposure to toxins, and infection are known to result in elevated levels of inflammation [6] and oxidative stress, both of which can increase a person’s risk of developing AD and other neurodegenerative disorders [4–6].

Herbicides are a ubiquitous part of our environment that may consequently pose harm to human health. Glyphosate (N-(phosphonomethyl) glycine) is the most heavily-applied herbicide in the US [7]. Approximately 300 million pounds are used annually in agricultural communities throughout the US [8]. Glyphosate works by inhibiting the enzyme enolpyruvylshikimate-3-phosphate in the shikimate pathway of plants preventing the production of aromatic amino acids critical for survival [9]. As of 2020, the US Environmental Protection Agency (EPA) stated that glyphosate poses no risks of concern to human health. However, the World Health Organization’s International Agency for Research on Cancer classified the herbicide as “possibly carcinogenic to humans” [10]. More recent research has highlighted that cancer may not be the only disease linked to glyphosate exposure [11], with multiple reports deciphering the effects of glyphosate on various organs and tissues, including the brain [6, 12, 13]. A recent publication from our group demonstrated that glyphosate could cross the blood brain barrier (BBB), being detected in brain tissue of exposed mice, and can elevate pro-inflammatory cytokines such as tumor necrosis factor-⍺ (TNF-⍺) [6]. Additionally, follow-up work from others has identified that glyphosate can, via elevated pro-inflammatory cytokines, disrupt mechanisms important for synaptic strengthening such as long-term potentiation [13]. Notably, a recent report found that 83.87% of humans included in the NHANES study had detectable urinary levels of glyphosate, with higher levels associated with decreased cognition [14]. Collectively, these findings suggest that there is an urgent need to understand the role of this herbicide in brain-related dysfunction.

Recent work demonstrates that glyphosate exposure may contribute to neuronal dysfunction, particularly via mechanisms linked to AD. In a recent study, a positive association was observed between urinary glyphosate and blood serum neurofilament light chain (NfL) levels [15]– NfL is a marker of neuronal axon damage, indicating that higher levels of glyphosate may be linked to neuronal damage [16]. Notably, the association was more pronounced in participants > 40 years of age and with a body mass index between 25 and 30 – subgroups with an increased risk for AD [1]. NfL is a blood biomarker that predicts cognitive decline across the AD continuum [16]. Additionally, previously published work demonstrated that glyphosate exposure to primary cortical neurons from the APP/PS1 mouse model of AD resulted in increased soluble Aβ levels [6]. These findings collectively highlight that more work is needed to determine the impact of glyphosate exposure on mechanisms associated with AD.

Due to the lack of a shikimate pathway in mammals, it has been suggested that ingested glyphosate is eliminated virtually unmetabolized [17–19]. However, recent studies have found that the gut microbiota can metabolize glyphosate into its major metabolite, aminomethylphosphonic acid [20, 21]. Additionally, chronic neuroinflammation can significantly alter cellular mechanisms important for behaviors such as learning and memory [22] and may contribute to neurodegenerative diseases where neuroinflammation is rampant [23, 24]. Given this, an important question that remains to be investigated is whether glyphosate and aminomethylphosphonic acid remain in the brain at detectable levels months after exposure has ceased. Evidence from prenatal exposure studies highlights that glyphosate exposure results in offspring with developmental delays, craniofacial abnormalities and cognitive deficits later in life, illustrating that early exposure can produce long-lasting consequences [25–27]. It remains to be known whether (1) glyphosate and aminomethylphosphonic acid remain in the brain after a recovery period and/or (2) an exposure period during early adulthood produces long-term detrimental consequences. Given that glyphosate and aminomethylphosphonic acid can cross the BBB [6, 28], it needs to be determined whether these molecules can remain in the brain and accelerate pathologies such as neuroinflammation, Aβ, and tau pathology.

The goal of the current study was to determine the impact of glyphosate exposure for 13 weeks, starting in early adulthood, followed by a six-month recovery period. Specifically, we sought to determine whether glyphosate and aminomethylphosphonic acid remained in the brain, and hypothesized that despite recovery from glyphosate exposure, neuroinflammation and exacerbation of AD-like neuropathologies – and associated cognitive deficits – would be observed in the 3xTg-AD mouse model of AD.

Methods

Animals and study design

3xTg-AD homozygous mice were generated as previously described [29, 30]. C57BL6/129S6 mice were used as non‐transgenic controls (NonTg). Notably, only female 3xTg‐AD mice were used because males do not display consistent neuropathology [30, 31], and the use of males is cautioned against [32], consistent with published work using this model [30, 33–35]. Mice were kept on a 12‐h light/dark cycle at 23 °C with ad libitum access to food and water and group‐housed, 4–5 per cage. Mice were randomly assigned to one of three doses of glyphosate starting at 4.5 months of age. The start age of glyphosate exposure was selected to begin prior to the presence of Aβ plaques and tau pathogenesis, pathologies which are detectable in 3xTg-AD mice at 6 months of age [30, 33]. Dosing ceased at ~ 7.5 months of age, for a total of 13 weeks, which is classified as chronic exposure consistent with previous publications [12, 36]. All animals were aged to 12 months for behavioral testing (4.5 months after exposure), as Aβ and tau pathogenesis is extensive in cortical and hippocampal brain regions at this age in 3xTg-AD mice [30, 33], allowing us to assess if early glyphosate exposure exacerbated cognitive deficits, Aβ, and tau pathogenesis. Blood was collected from mice at 13.5 months of age, and all animals were subsequently euthanized for tissue harvesting, when pathology is well advanced in the 3xTg-AD mouse [33]. All animal procedures were approved in advance by the Institutional Animal Care and Use Committee of Arizona State University, protocol number 22-1933R.

Justification of glyphosate doses

As previously described by our group [6], chemically pure glyphosate (N-(Phosphonomethyl)glycine; C₃H₈NO₅P) was purchased from Sigma-Aldrich (product number P9556) and prepared at 0.107 g/L in 1.89 M sodium hydroxide. The stock solution was serially diluted with deionized (DI) water to the desired doses and adjusted to a pH of 7, with the same solution without glyphosate serving as a vehicle. Mice were randomly assigned to receive one of three doses starting at 4.5 months of age: vehicle (0 mg/kg), 50 mg/kg, or 500 mg/kg of body weight. Mice were dosed daily over the course of the 13-week exposure period. The high dose (500 mg/kg) was based on the no observable adverse effect limit (NOAEL) for chronic (90 day) exposure in mice established by the EPA [37], and has been used in publications to determine the effects of glyphosate exposure on neurocognitive and peripheral organ alterations [12, 36, 38, 39]. The lower dose of 50 mg/kg was selected as it is 10 fold less than the NOAEL [37], and is less than the value that was used to calculate the chronic reference dose for humans of 1.75 mg/kg/day [37], which was 175 mg/kg [40]. This allows us to compare our results to previous published work assessing glyphosate exposure and neurocognitive effects [12, 36, 38]. The number of mice per group was 0 mg/kg (n = 14 3xTg-AD; n = 15 NonTg), 50 mg/kg (n = 13 3xTg-AD; n = 15 NonTg), and 500 mg/kg (n = 15/group for both 3xTg-AD and NonTg). The number of mice used for each analysis was similar to previous published work [31, 33, 41].

Behavior testing

At 12 months of age, all mice were tested in the Morris water maze (MWM) task to assess hippocampal-dependent spatial learning and memory, as previously described [42, 43]. All animals underwent 4 training trials/day for 5 days. The location of the hidden platform remained constant, but the start location pseudo‐randomly varied across trials. Mice were given 60 s/trial to locate the hidden platform. Twenty‐four hours after the last training session, the platform was removed, and mice were returned to the MWM for 60 s to assess spatial reference memory. Data were collected with EthoVisionXT (Noldus Information Technology).

Blood collection and plasma extraction

Blood was collected via the submandibular vein prior to euthanasia at 13.5 months of age. 150–200 µL (≤ 1% of the subject’s body weight) of blood was collected and placed into EDTA-lined tubes (BD K₂EDTA #365974) and inverted eight times to assure anticoagulation. Tubes were kept on ice for 60–90 min and then centrifuged at 455 g for 30 min at 4 °C to separate phases. The top layer was collected and frozen at − 80 °C.

Circulatory glyphosate and aminomethylphosphonic acid quantification

Glyphosate and aminomethylphosphonic acid were in a concentration range of 0–3 ng/mL in plasma, and quantified as follows: briefly, pooled lyophilized human plasma (Innovative research, Novi, MI) was used for preparing calibration curves and quality controls by spiking variable concentrations of glyphosate and aminomethylphosphonic acid. Calibration curves, quality controls and mouse plasma (5–20 µL) spiked with internal standards (¹³C¹⁵N glyphosate and ¹³C¹⁵N aminomethylphosphonic acid at 6.26 ng/mL concentration) were subjected to protein removal using 500 µL 10 kDa molecular weight cut off (MWCO, Millipore Sigma, Burlington, MA) spin columns (10,000 g, 20 min, 4 ℃). The filtrate was acidified with formic acid to a final concentration of 0.1% (v/v) and used for Liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis. Plasma measurements showed excellent linearity with an R² > 0.99 (Supplemental Fig. 1), and precision < 15% relative standard deviation (RSD), and accuracy > 87% for both glyphosate and aminomethylphosphonic acid. The observed lower limit of quantitation (LOD) and limit of quantitation (LOQ) were 10 pg/mL and 50 pg/mL for glyphosate and aminomethylphosphonic acid, respectively.

Tissue harvesting and processing, western blots, and enzyme-linked immunosorbent assay (ELISA)

Mice were euthanized at an average age of 13.5 months. All mice were perfused with fresh 1X Phosphate buffered saline (PBS) to remove blood from the brain. Brains were extracted and the left hemisphere was fixed in 4% paraformaldehyde for 48 h; from the right hemisphere, we isolated the entire cerebral cortex, then removed the hippocampus (Hp). The cortex (Ctx) fraction represents the cortical plate – including the isocortex, olfactory areas, and retrohippocampal cortex – and the cortical subplate. Dissected Ctx tissue was prioritized for glyphosate, aminomethylphosphonic acid and cytokine/chemokine measures, where Hp tissue was not, due to the limited amount of sample compared to the Ctx. The Hp and remaining Ctx were prioritized for Aβ and pathological tau measures, given that a major goal was to determine AD-like pathology exacerbation effects with early glyphosate exposure in key areas affected in AD. Hp and Ctx tissue were prepared for protein assays as previously described [42, 43]. Dissected tissue was homogenized in tissue protein extraction reagent supplemented with protease (Roche Applied Science, IN, USA) and phosphatase inhibitors (Millipore, MA, USA). The homogenized tissues were centrifuged at 21,130 g at 4 °C for 30 min, and the supernatant (soluble fraction) was stored at − 80 °C. We then homogenized the pellet in 70% formic acid followed by centrifuging at 4 ℃ for 30 min, to extract insoluble proteins. Western blots were performed under reducing conditions as previously described [44], using the following antibodies: Biolegend, 6E10 (Full length (FL)-amyloid precursor protein (APP), Catalog #9320-02, 1:1,000); Abcam, Glyceraldehyde 3-phosphate dehydrogenase (GAPDH, Catalog #ab8245, 1:3,000). Licor Image Studio software was used for quantitative analyses by normalizing the intensity of the protein of interest with its loading control GAPDH within each blot. Mean density values were normalized for each immunoblot by dividing each experimental band density by the mean 3xTg-AD 0 mg/kg density for the Hp and Ctx. The experimenter was blinded to the group allocations. Soluble fractions of Hp protein homogenates were probed for 99 amino acid C-terminal fragment of amyloid precursor protein (C99, MyBioSource catalog #MBS7612253) and Beta-secretase 1 (BACE-1, LSBio catalog# LS-F7271) using commercially available ELISA kits. Soluble and insoluble fractions of Hp and Ctx were probed for human Aβ₄₀ and Aβ_42, and phosphorylated tau (pTau) at Threonine 181 (Thr 181) and Serine 396 (Ser 396) residues, using the commercially available ELISA kits (Invitrogen-ThermoFisher Scientific catalog #KMB3481, KMB3441 and KHB7031, respectively) as previously described [42]. All samples were run in duplicate wells.

Brain glyphosate and aminomethylphosphonic acid measurements

Glyphosate and aminomethylphosphonic acid measurements were conducted as previously described [6]. Samples were spiked-in with 10 ng/g of ¹³C₂¹⁵N glyphosate (Toronto Research Chemicals) and ¹³C¹⁵N aminomethylphosphonic acid (Toronto Research Chemicals). Lipid removal was carried using 1 cc Sep-Pak C18 solid phase cartridges (Waters, Milford-MA) and the flow-through (400 µL) were acidified with formic acid. Samples, calibration curve standards and quality control standards in 50 or 100 µL injection volume were analyzed by liquid chromatography with multiple reaction monitoring (LC-MRM) on Vanquish Duo Ultra high-pressure liquid chromatography (UHPLC) system coupled to a Thermo TSQ Altis instrument [6, 45]. Calibration curves performed in the mouse brain matrix over a range of 0–60 ng/g of glyphosate and aminomethylphosphonic acid showed excellent linearity (R² > 0.99, Supplementary Fig. 2). The observed LOQ for the assay defined as the lowest spiked-in standard with a mean accuracy between 80 and 120% and precision less than 20% RSD was 0.4 ng/g for glyphosate and aminomethylphosphonic acid respectively [6, 46].

Multiplex cytokine assay

The long-lasting inflammatory effects of glyphosate were tested in blood plasma and cortical protein homogenates using a Bio-Plex mouse cytokine 23-plex kit (Bio-Rad, Catalog #M60009RDPD). Briefly, 15 µl of plasma were diluted in 35 µl diluent solution per assay instructions. Soluble cortical protein homogenates containing 100 µg total protein per well were added in duplicate to a 96-well plate and assayed according to manufacturer’s instructions. Measurements were calculated with the Bio-Plex Manager software. Standard curves were plotted using five-parameter logistic regression and concentrations were calculated accordingly. Levels that were not detected based on manufacture limits were not included in the statistical analysis.

Tissue sectioning, histology and ImageJ analysis

Brain hemispheres were sectioned into 50 μm coronal sections using a vibratome and stored in 1X PBS with 0.02% sodium azide. For Aβ₄₂ staining, one section per animal including the ventral extent of the Hp underwent immunohistochemistry as previously described [44], using an antibody to stain for Aβ₄₂ plaques (anti-Aβ₄₂, 1:200 dilution, Millipore, catalog# 5078P). Sections were permeabilized in 88% formic acid for 7 min before incubation in primary antibody. Brightfield photomicrographs were taken at 5x with a Zeiss Axio Imager, stitched together in Adobe photoshop and analyzed via ImageJ analyze particle function at a threshold of 0/70 as previously described [47]. For Thioflavin S staining, four sections per animal including the ventral extent of the Hp were included for analysis. Tissue sections were incubated in 4% paraformaldehyde for 15 min, filtered 1% aqueous Thioflavin S for 10 min at room temperature, washed twice in 80% ethanol, once in 95% ethanol, and 3 times in double distilled H2O. Images were taken at 5x on a fluorescence microscope (Leica DMi8) with Leica Application Suite X (LAS X) software, and quantified using ImageJ for % area of particles in the Hp as previously described [42, 48]. AT8 immunohistochemistry was performed as previously described [42, 43] with an antibody to stain cells expressing tau at Ser 202, Thr 205 (1:500 dilution, Invitrogen, catalog# MN1020). A series of coronal tissue sections including the dorsal and ventral extent of the Hp were evaluated for AT8, rendering seven sections per animal.

Unbiased stereology for AT8 + neuron quantification

Stereoinvestigator software V17 (Micro-BrightField, Cochester, VT) optical fractionator method was used to quantify AT8 + cornu ammonis 1 (CA1) cells in the Hp as previously described [42, 43]. Counts were performed at predetermined intervals; grid size (X and Y = 158 μm), counting frame (X and Y = 50 μm), superimposed on the live image of the tissue sections. Coronal tissue sections were analyzed using a 63x × 1.4 PlanApo oil immersion objective. Gunderson’s scores remained ≤ 0.07. The average tissue thickness was 18.6 μm. Dissector height was set at 15 μm, with a 2-µm top and 2-µm bottom guard zone. Bright-field photomicrographs were taken with a Zeiss Axio Imager. The AT8 antibody penetrated the full depth of the section, allowing for an equal probability of counting all objects.

Statistical analysis

Two-way factorial Analysis of variance (ANOVA; for genotype and dose) was used to analyze experimental data, followed by recommended corrected post hoc tests when appropriate using Graph Pad PRISM (Version 10). Repeated measures ANOVA was used to analyze the MWM data output. One-way ANOVA was utilized for comparison of 3xTg‐AD mice at the three doses for neuropathological measures solely present in humanized mice. Violation of homogeneity of variance was followed by non-parametric analysis. Statistical outliers were identified using the ROUT and Grubbs method. Significance was set at p < 0.05.

Results

Glyphosate exposure for 13 weeks followed by a 6-month recovery does not alter body weight but increases incidence of premature death

To determine whether exposure to glyphosate followed by recovery results in exacerbated neuroinflammation, neuropathology, and associated cognitive deficits, we exposed NonTg and 3xTg-AD mice to either 0 mg/kg (n = 14 3xTg-AD; n = 15 NonTg), 50 mg/kg (n = 13 3xTg-AD; n = 15 NonTg), or 500 mg/kg/day (n = 15/group) of glyphosate via oral gavage for 13 weeks starting at 4.5 months of age. At ~ 7.5 months of age, for the recovery period, dosing ceased for the remainder of life (Fig. 1A). The body weight of mice during the dosing period did not change (Fig. 1B, C). During the recovery period, between 7.5 and 13.5 months of age, and consistent with published work [30, 42], we found a significant main effect of genotype in % weight change, where the 3xTg-AD mice gained significantly more weight than the NonTg counterparts (F_(1,70) = 7.914, p = 0.0064; Fig. 1D). Lastly, we examined premature deaths during the dosing period and found that n = 1 NonTg 0 mg/kg and n = 1 NonTg 500 mg/kg mice died (Fig. 1E). The remaining NonTg mice survived the remainder of the study. During the dosing period, n = 1 3xTg-AD 0 mg/kg and n = 2 3xTg-AD 500 mg/kg mice died, and during the recovery period, n = 4 3xTg-AD 50 mg/kg and n = 2 3xTg-AD 500 mg/kg mice died (Fig. 1F, G). Collectively, these results highlight that exposure to glyphosate does not affect body weight but contributed to premature death, in particular in the 3xTg-AD mice.

Fig. 1.

Glyphosate exposure followed by a recovery period does not impact body weight but reduces survival in 3xTg-AD mice. (A) Experimental design. (B) Body weight across the dosing period. (C, D) % weight change during the dosing period, and % weight change during the recovery phase (end of dosing to end of study). (E, F) Survival curves during the dosing and recovery period. (G) Summary table of the number of mice at the start and conclusion of the study. Line and bar graphs are means ± SEM. **p < 0.01

Glyphosate exposure increases thigmotaxia in the Morris water maze task in 3xTg-AD mice

To determine whether glyphosate exposure impaired spatial cognition, mice were tested in the MWM for 6 consecutive days (0 mg/kg n = 13 3xTg-AD, n = 14 NonTg; 50 mg/kg n = 12 3xTg-AD, n = 15 NonTg; 500 mg/kg n = 12 3xTg-AD, n = 14 NonTg). During the first 5 training days, mice received 4 trials/day. We found a significant main effect of day in distance traveled to find the hidden platform, indicating learning (F_{(4, 74)} = 35.884, p < 0.0001; Fig. 2A). We also found a significant main effect of genotype (F_{(1, 74)} = 16.670, p = 0.0001), where 3xTg-AD mice traveled significantly further to find the hidden platform than NonTg mice. No significant dose main effects or interactions were found for distance. We next performed analysis of % thigmotaxia, the tendency to remain near walls. Notably, thigmotaxia has been identified as an indicator of anxiety [49]. During the learning phase, we found a significant effect of day (F_{(1, 74)} = 49.490, p < 0.0001; Fig. 2B), indicating that mice showed reduced % thigmotaxia across days. We also found a significant genotype by dose interaction (F_{(2, 74)} = 3.463, p = 0.0365). Post hoc analysis revealed that the 3xTg-AD 500 mg/kg group exhibited more thigmotaxia during the learning phase of the MWM compared to the 3xTg-AD 0 mg/kg group (p = 0.0123), indicating increased anxiety. On Day 6, the platform was removed, and mice were tested in a 60 s probe trial to assess spatial memory. We found a significant main effect of genotype (F_{(1, 74)} = 4.043, p = 0.048; Fig. 2C), where the 3xTg-AD mice crossed the platform location significantly less than the NonTg counterparts. We also found a significant main effect of genotype (F_{(1, 74)} = 9.778, p = 0.0025; Fig. 2D) for swim speed, where the 3xTg-AD mice showed increased velocity compared with the NonTg mice, indicating that any deficits in performance were not a result of impaired swim speed. These data indicate that 3xTg-AD mice exposed to glyphosate showed increased thigmotaxia, evident of exacerbated anxiety-like behavior, although overall performance was equal across the dosed groups.

Fig. 2.

Glyphosate exposure followed by a recovery period leads to the presence of aminomethylphosphonic acid in cortical tissue and increases thigmotaxia in the morris water maze (MWM). (A-D) Performance in the learning and memory phases of the MWM. (E, F) Glyphosate and aminomethylphosphonic acid in blood plasma of mice after the recovery period. (G) Brain weights of mice at the end of the study. (H) Aminomethylphosphonic acid was present months after glyphosate exposure in cortical brain tissue. Line and bar graphs are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Aminomethylphosphonic acid is detected in brain tissue despite months of glyphosate recovery

To determine whether glyphosate and aminomethylphosphonic acid remain present in blood and brain tissue months after exposure, we performed UPLC-MS on blood plasma and cortical tissue homogenates of mice (n = 5–6/group). In blood plasma, while both glyphosate and aminomethylphosphonic acid were detected, likely resulting from glyphosate presence in chow, no significant differences between the groups were observed (Fig. 2E, F). When examining brain weight, we found a significant main effect of genotype, where the 3xTg-AD mice show significant reductions compared to the NonTg mice (F_(1,68) = 51.83, p < 0.0001; Fig. 2G), consistent with published work [30]. We also found a significant main effect of dose (F_{(2, 68)} = 3.607, p = 0.0325), where the 50 mg/kg dose had lower brain weight than the 0 mg/kg group (p = 0.0277). We did not detect glyphosate in cortical tissue. Interestingly, we found a significant main effect of dose for cortical aminomethylphosphonic acid levels, where the 50- and 500 mg/kg groups had significantly higher levels than the 0 mg/kg regardless of genotype (F_(2,26) = 20.81, p < 0.0001; Fig. 2H). Collectively, these results highlight that 6 months after glyphosate exposure, brain weight is reduced in the 50 mg/kg groups and cortical aminomethylphosphonic acid is significantly elevated in mice at both 50 mg/kg and 500 mg/kg groups, highlighting lasting effects.

3xTg-AD glyphosate exposed mice show increased amyloidogenic processing and increased soluble and insoluble fractions of Aβ and plaques in the hippocampus (Hp) and cortex (Ctx)

Mice were euthanized at 13.5 months of age and brains were collected for neuropathological assessment. Given previous reports showing that glyphosate exposure increases amyloidosis in vitro [6], we first examined whether glyphosate exposure in vivo altered levels of full length human APP (n = 4/group) and the C99 fragment of APP (n = 6/group), which is cleaved to form the Aβ peptide. NonTg mice do not display humanized Aβ pathology and therefore were excluded from these analyses [42, 44]. We found no significant increase in steady state levels of APP in either Hp or Ctx tissue (Fig. 3A, B). For C99 in the Hp, we found a significant dose-dependent effect (F_(2,15) = 62.26, p < 0.0001; Fig. 3C), where 3xTg-AD 50 mg/kg (p < 0.0001) and 500 mg/kg (p < 0.001) mice showed higher levels than 0 mg/kg, and 500 mg/kg exhibited higher levels than the 50 mg/kg mice (p = 0.0020). Next, we examined the levels of BACE-1 (n = 6/group), the initiator enzyme that cleaves the transmembrane APP towards the amyloidogenic pathway [50]. We found a significant effect (F_{(2, 15)} = 6.438, p = 0.0096; Fig. 3D), where 3xTg-AD 50 mg/kg (p = 0.0347) and 500 mg/kg (p = 0.0061) mice showed a significant increase compared to the 3xTg-AD 0 mg/kg group, collectively highlighting an increase in the enzyme and cleaved product of the amyloidogenic pathway after the recovery period of glyphosate exposure.

Fig. 3.

Glyphosate exposure increases soluble and insoluble fractions of Aβ40 and 42, and Aβ42 plaque load by increasing cleavage products of the amyloid precursor protein despite a recovery period. (A, B) Representative western blot and quantification of full length APP (6e10) with loading control GAPDH in hippocampal (Hp) and cortex (Ctx) tissue. (C, D) Glyphosate exposure increases C99 and BACE-1 protein levels probed after the recovery period. (E-H) Soluble and insoluble fractions of Aβ40 and 42 in Hp and Ctx tissue. (I, J) Photomicrographs showing significant increases in both % Aβ42 plaque load in the Ctx of 3xTg-AD 500 mg/kg mice, and in average particle size in the Hp of 3xTg-AD 50 mg/kg mice. (L, M) Photomicrographs depicting Thioflavin S structures and quantification showing a trending increase in % particle area in glyphosate exposed 3xTg-AD mice (p = 0.0659) . Abbreviations: DS = dorsal subiculum, DG = dentate gyrus, CA1 = cornus ammonis 1. Bar graphs are means ± SEM. For box plots, the center line represents the median value, the limits represent the 25th and 75th percentile, and the whiskers represent the minimum and maximum value of the distribution.*p < 0.05, **p < 0.01, ****p < 0.0001

To understand the effects of glyphosate exposure on AD pathogenesis, we used ELISAs to quantify soluble and insoluble Aβ₄₀ and Aβ₄₂ fractions in the 3xTg-AD mice (n = 5–7/group). For soluble Aβ₄₀ fractions, we found no significant differences in the Hp and Ctx of 3xTg-AD mice across the doses (Fig. 3E). For insoluble Aβ₄₀ fractions, we found a significant effect (F_{(2, 15)} = 5.305, p = 0.0181; Fig. 3F), where the Hp of 3xTg-AD 500 mg/kg mice showed higher levels than the 3xTg-AD 0 mg/kg (p = 0.0177) mice, but no significant differences were detected in the Ctx. Next, we examined the soluble fractions of Aβ₄₂ and found significant effects in the Hp (F_{(2, 15)} = 9.726, p = 0.0020; Fig. 3G), where the 500 mg/kg 3xTg-AD mice showed higher levels compared to the 0 mg/kg (p = 0.0107) and 50 mg/kg (p = 0.0028) groups. No differences in the Ctx were detected for soluble Aβ₄₂. For insoluble Aβ_42, we found a significant dose-dependent effect in the Hp (F_(2,15) = 75.06, p < 0.0001; Fig. 3H), where the 3xTg-AD 500 mg/kg exhibited higher levels than the 3xTg-AD 50 mg/kg (p < 0.0001), and 3xTg-AD 50 mg/kg showed higher levels than the 3xTg-AD 0 mg/kg mice (p = 0.0418). Similarly, we found a significant dose-dependent effect of insoluble Aβ₄₂ in the Ctx (F_{(2, 14)} = 46.58, p < 0.0001), where the 3xTg-AD 500 mg/kg exhibited higher levels than the 3xTg-AD 50 mg/kg (p = 0.0137), and 3xTg-AD 50 mg/kg showed higher levels than the 3xTg-AD 0 mg/kg mice (p < 0.0001). Aβ₄₂ is more prone to aggregation and toxicity than Aβ₄₀ [51], thus we next quantified Aβ₄₂ plaque load and plaque (particle) size (n = 5–6/group). We did not find any differences in the Hp for % change in Aβ₄₂ plaque load, however, we did find a significant effect of dose in the Ctx (F_(2,14) = 4.675, p = 0.0279; Fig. 3I, J), where the 3xTg-AD 500 mg/kg mice exhibited a higher load than the 50 mg/kg and 0 mg/kg counterparts (p = 0.0440). We also found a significant effect in the Hp for plaque size (F_{(2, 14)} = 6.126, p = 0.0123; Fig. 3K), where the 3xTg-AD 50 mg/kg dosed mice had larger-sized plaques than the 0 mg/kg and 500 mg/kg counterparts. We found no significant effects in the Ctx for particle size. Lastly, we stained tissue for Thioflavin S to verify the extent of β-sheet confirmation aggregates in the ventral Hp of 3xTg-AD mice (n = 5–6/group). We found a trending increase in the % area of particles in the Hp of 50 mg/kg and 500 mg/kg groups compared to 0 mg/kg groups X²₍₂₎ = 5.440, p = 0.0659 (Fig. 3L, M). Collectively, these results demonstrate that early glyphosate exposure alters APP processing, thereby exacerbating Aβ levels in key areas affected in AD, with most significant differences evident in insoluble Aβ₄₂ fractions, and Aβ₄₂ plaque load and plaque size, despite a 6-month recovery.

3xTg-AD glyphosate-exposed mice exhibit significantly increased pathological tau in the Hp and Ctx

We next sought to understand the long-lasting effects of glyphosate exposure on tau pathogenesis. To accomplish this, we performed ELISAs to detect soluble and insoluble fractions of phosphorylated tau (pTau) at Thr 181 and Ser 396 in 3xTg-AD mice (n = 6/group). For soluble pTau Thr 181, we found a significant effect in the Hp (F_{(2, 15)} = 25.54, p < 0.0001; Fig. 4A), where both the 3xTg-AD 500 mg/kg (p < 0.0001) and 50 mg/kg (p < 0.0001) mice exhibited higher levels than the 3xTg-AD 0 mg/kg group. In the Ctx, we found a significant dose-dependent effect (F_{(2, 15)} = 926.80, p < 0.0001), where the 3xTg-AD 500 mg/kg exhibited higher levels than the 3xTg-AD 50 mg/kg (p < 0.0001), and 3xTg-AD 50 mg/kg showed higher levels than the 3xTg-AD 0 mg/kg mice (p < 0.0001). For soluble pTau Ser 396, we found significant dose-dependent effects in the Hp (F_{(2, 15)} = 28.12; Fig. 4B) and Ctx (F_{(2, 15)} = 79.37, p < 0.0001), where the 3xTg-AD 500 mg/kg mice exhibited higher levels than the 3xTg-AD 50 mg/kg mice (Hp p = 0.0046, Ctx p < 0.0001; Fig. 4B), and 3xTg-AD 50 mg/kg mice showed higher levels than the 3xTg-AD 0 mg/kg mice (Hp p = 0.0073, Ctx p < 0.0001). For insoluble pTau Thr 181, we found a significant effect of dose for the Hp (F_{(2, 15)} = 24.61, p < 0.0001; Fig. 4C) and Ctx (F_{(2, 15)} = 6.600, p = 0.0088), where the 3xTg-AD 50 mg/kg group showed higher levels in the Hp compared to the 0 mg/kg (p < 0.0001) and 500 mg/kg (p = 0.0005) counterparts, and the 3xTg-AD 500 mg/kg had higher levels than the 0 mg/kg (p = 0.0084) in the Ctx. For insoluble pTau Ser 396, we found a significant effect in the Hp (F_{(2, 15)} = 7.025, p = 0.0070; Fig. 4D), where the 3xTg-AD 500 mg/kg mice showed higher levels than the 3xTg-AD 0 mg/kg group (p = 0.0065). In the Ctx, we found a significant effect (F_{(2, 15)} = 26.26, p < 0.0001), where the 3xTg-AD 500 mg/kg showed higher levels than the 3xTg-AD 50 mg/kg (p = 0.0010) and 3xTg-AD 0 mg/kg (p < 0.0001) groups. Lastly, we stained tissue against AT8 (n = 6/group), which is associated with intraneuronal tau filaments [43], and found significantly higher AT8 + cells in CA1 of the Hp of 3xTg-AD 50 mg/kg and 500 mg/kg compared to the 0 mg/kg (t₍₆₎ = 2.462, p = 0.049; Fig. 4E, F). These results are the first demonstration that glyphosate exposure increases hyperphosphorylated tau. Collectively, these data demonstrate that early exposure to glyphosate increases AD-like neuropathology in 3xTg-AD mice.

Fig. 4.

Glyphosate exposure followed by a recovery period increases phosphorylated pathological tau. (A-D) Soluble and insoluble fractions of phosphorylated tau at Threonine (Thr) 181 and Serine (Ser) 396 in hippocampal (Hp) and cortex (Ctx) tissue. (E, F) AT8 (Ser 202/Thr 205) + cell number was significantly elevated in the CA1 region of the Hp of 3xTg-AD 50 mg/kg and 500 mg/kg. Abbreviations: DS = dorsal subiculum, DG = dentate gyrus, CA1 = cornus ammonis 1. Bar graphs are means ± SE. For box plots, the center line represents the median value, the limits represent the 25th and 75th percentile, and the whiskers represent the minimum and maximum value of the distribution. Bar graphs are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

3xTg-AD glyphosate-exposed mice exhibit significantly increased cytokines and chemokine levels in blood plasma

Elevation of cytokines (pro- and anti-inflammatory) and chemokines (secreted proteins within the cytokine family that induce cell migration) is indicative of inflammation [52]. Previous work shows that glyphosate exposure increases the levels of pro-inflammatory cytokines in circulating blood, liver, and the brain, such as TNF-α, which has been linked to neurotoxicity, cell death and disorders such as AD [6, 53, 54]. To determine whether early glyphosate exposure produced long-lasting peripheral inflammatory changes, we performed a 23-plex cytokine/chemokine panel of peripheral blood plasma of mice (n = 7–9 NonTg; n = 3–7/ 3xTg-AD). Levels of cytokines and chemokines were non-detectable in NonTg blood plasma. However, we did find a significant genotype by dose interaction for all 23 markers, indicating significant elevations in the 3xTg-AD-dosed mice (Fig. 5A). Figure 5A highlights the significant elevations across all 23 markers, suggesting that despite a 6-month recovery, glyphosate exposure induced long-lasting increases in peripheral inflammatory molecules that may have contributed to the increased mortality in 3xTg-AD mice.

Fig. 5.

Glyphosate exposure followed by a recovery period increases cytokine and chemokine levels in blood plasma in 3xTg-AD mice, and in both NonTg and 3xTg-AD cortical tissue. (A, B) Heat map illustrating significant increases in cytokine and chemokines in blood plasma in 3xTg-AD mice (values were below detection levels in NonTg mice), and in cortical tissue homogenates of both NonTg and 3xTg-AD mice. (C-I) Significant genotype by dose interactions were graphed to illustrate differences across the two variables. Bar graphs are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Glyphosate-exposed NonTg and 3xTg-AD mice show increases in cytokines and chemokine levels in the Ctx

Neuroinflammation is a critical pathophysiological process implicated in the development of neurodegenerative diseases [23, 24]. To determine whether early glyphosate exposure produced long-lasting inflammatory changes in the brain, we performed the 23-plex panel using cortical brain homogenates of a subgroup of mice (n = 4/group). We found significant main effects of genotype and dose for all 23 markers, with 3xTg-AD showing higher levels than NonTg and the 50 mg/kg and 500 mg/kg showing higher levels than 0 mg/kg groups (Fig. 5B). We also found a significant genotype by dose interactions for seven pro-inflammatory cytokines. For Eotaxin (C-C motif ligand 1, CCL1), an immune marker of accelerated aging and neurodegeneration [55], we found a significant genotype by dose interaction (F_{(2, 18)} = 5.987, p = 0.0102; Fig. 5C), where NonTg 500 mg/kg mice showed higher levels than 50 mg/kg (p = 0.0017) and 0 mg/kg (p < 0.0001) NonTg counterparts. For granulocyte-colony stimulating factor (G-CSF), a growth factor that promotes neuroprotective effects and is dysregulated in AD [56], we found a significant genotype by dose interaction (F_{(2, 17)} = 4.416, p = 0.0286; Fig. 5D), where the NonTg 500 mg/kg mice showed higher levels than 50 mg/kg (p = 0.0170) and 0 mg/kg (0.0012) NonTg counterparts. For interferon-gamma (IFN-γ), a pro-inflammatory cytokine that activates macrophages and is an inducer of class II major histocompatibility complex (MHC), whose presence in the brain has been associated with breach of the BBB and neurodegeneration [57, 58], we found a significant genotype by dose interaction (F_{(2, 18)} = 4.792, p = 0.0215; Fig. 5E). Post hoc analysis revealed a significant dose-dependent effect in the NonTg mice, where the 500 mg/kg had higher levels than the 50 mg/kg group (p < 0.0001), and the 50 mg/kg group had higher levels than the 0 mg/kg groups (p = 0.0455). In the 3xTg-AD mice, we found a significant elevation in the 50 mg/kg mice compared to the 0 mg/kg group (p = 0.0087). For interleukin (IL) -9, a pro-inflammatory cytokine that promotes cell proliferation, prevents apoptosis, and is associated with progression of mild cognitive impairment (MCI) to AD [59], we found a significant genotype by dose interaction (F_{(2, 18)} = 4.872, p = 0.0204; Fig. 5F). Post hoc analysis revealed that NonTg 500 mg/kg had significantly higher levels than both the NonTg 50 mg/kg group (p = 0.0049) and 0 mg/kg group (p < 0.0001). For IL-10, an anti-inflammatory cytokine that limits immune response to maintain homeostasis and is elevated in AD [60, 61], a significant genotype by dose interaction revealed (F_(2,17) = 7.745, p = 0.0041; Fig. 5G) elevations in the NonTg 500 mg/kg group compared to the 50 mg/kg (p = 0.0003) and 0 mg/kg NonTg groups (p < 0.0001). 3xTg-AD 50 mg/kg dose mice showed significant elevations compared to the 0 mg/kg counterparts (p = 0.0441). The chemokine keratinocyte-derived cytokine (KC), also known as CXCL1, is upregulated in response to inflammation, however it has been shown to increase tau hyperphosphorylation, a pathogenic mechanism observed in AD [62]. We found a significant genotype by dose interaction (F_{(2, 17)} = 6.723, p = 0.0071; Fig. 5H), where NonTg 500 mg/kg showed an increase compared to the 50 mg/kg (p = 0.0001) and 0 mg/kg (p < 0.0001) NonTg groups. In 3xTg-AD mice, we found a significant increase in the 50 mg/kg (p = 0.0182) and 500 mg/kg (p = 0.0085) compared to the 0 mg/kg group. Lastly, we measured levels of macrophage inflammatory protein (MIP) 1-β, also known as CCL4, which has been shown to disrupt neurovascular endothelium and increase Aβ [63, 64]. We found a significant genotype by dose interaction (F_{(2, 18)} = 5.598, p = 0.0129; Fig. 5I), where the NonTg 500 mg/kg group had significantly higher levels than the 50 mg/kg (p = 0.0015) and 0 mg/kg (p < 0.0001) NonTg counterparts. Collectively, these results highlight that even after a 6-month recovery, inflammatory cytokines and chemokines are elevated in cortical tissue of both NonTg and 3xTg-AD mice, and may contribute to the progression of AD-pathogenesis.

Discussion

In the present study, we sought to determine whether glyphosate and aminomethylphosphonic acid would persist in mice following a 6-month recovery period, and whether this exposure/recovery paradigm would alter neuroinflammation and AD pathology. Aminomethylphosphonic acid was detectable in brain tissue of mice dosed with glyphosate, with higher-dosed animals exhibiting higher accumulation. Additionally, we found a significant increase in soluble and insoluble fractions of Aβ, Aβ₄₂ plaque load and plaque size, and pTau at Thr 181 and Ser 396 in Hp and Ctx brain tissue from glyphosate-exposed mice, highlighting an exacerbation of hallmark AD-like proteinopathies. We also observed elevations in AT8 in the Hp, collectively demonstrating increased tau pathogenesis with glyphosate exposure for the first time. Notably, we found elevations in pro- and anti-inflammatory cytokines and chemokines in blood plasma of 3xTg-AD mice and in both 3xTg-AD and NonTg mice cortical brain tissue despite recovery. This is significant as prolonged inflammation has been shown to affect the progression of AD pathology and increase the likelihood of developing a neurodegenerative disease [52, 65–67]. Thigmotaxia - a behavior where animals spend more time close to walls and one which is classified as an anxiety-like behavior [68] - was significantly elevated in 3xTg-AD glyphosate-exposed mice. Taken together, our results are the first to show increased AD-like pathology and elevated inflammation in the peripheral blood and brains of glyphosate-exposed mice, with presence of aminomethylphosphonic acid months after glyphosate exposure (Fig. 6).

Fig. 6.

Graphical abstract of our major findings highlighting that exposure to glyphosate followed by a significant recovery period was capable of eliciting long-lasting pathological consequences

Mice dosed with glyphosate did not show significant weight loss during the dosing and recovery periods of the study, consistent with previous reports in glyphosate exposed rodents [18, 69, 70]. However, a higher percentage of 3xTg-AD mice died prematurely during the dosing and recovery period, suggesting that underlying health issues in the AD model may be exacerbated by glyphosate exposure. For example, 3xTg-AD mice exhibit impaired glucose metabolism and insulin resistance [29], as well as an increased susceptibility to high inflammatory responses following acute infection when compared to NonTg mice [71]. Here, we found that elevated peripheral cytokine and chemokine levels were present in 3xTg-AD mice months after glyphosate recovery, but not in NonTg mice. While we did not assess glucose levels or organ pathologies beyond peripheral blood and brain tissue, it is possible that the increased peripheral blood cytokine and chemokine levels observed in 3xTg-AD mice contributed to an exacerbation of already-present ailments, leading to decreased survival. Previous work in rats has shown that glyphosate exposure disrupted glucose homeostasis and insulin signaling, leading to hepatic inflammation and liver damage – all factors which can lead to type II diabetes [72]. Additionally, work in humans has shown that higher urine glyphosate concentrations are associated with type II diabetes, hypertension, cardiovascular disease, and obesity [73]. Thus, it is tempting to speculate that glyphosate exposure may have increased such metabolic ailments, contributing to premature death in 3xTg-AD mice. Future work is needed to dissect such interactions.

Aminomethylphosphonic acid shows similar toxicity to glyphosate [74], and previous work has identified that measurable aminomethylphosphonic acid in humans is associated with oxidative damage and adverse metabolic outcomes [54, 75]. Surprisingly, we found that aminomethylphosphonic acid was present in brain tissue of both 3xTg-AD and NonTg mice months after glyphosate dosing. The mechanism by which aminomethylphosphonic acid entered and was present in cortical brain tissue after glyphosate exposure remains unknown. The most probable hypothesis is that, during the dosing regimen, glyphosate was metabolized into aminomethylphosphonic acid by the gut microbiome. This is supported by evidence that the gut microbiome of rodents and humans contain microbes, such as Lactobacilli and Bacteroides, with a shikimate pathway. Most of the bacteria in the human gut microbiome contain the class of 5-enolpyruvylshikimate 3-phosphate synthase enzymes rendering them sensitive to inhibition by glyphosate [21, 76]. Subsequently, we hypothesize that aminomethylphosphonic acid crossed the BBB, leading to its detection in cortical tissue months after glyphosate exposure. This is supported by work showing that aminomethylphosphonic acid crosses the BBB in an in vitro model [28]. Furthermore, 3xTg-AD mice show microvascular degeneration that is evident prior to plaque onset and gets worse with age [77], perhaps due to an impaired capacity to maintain BBB integrity which has been shown using in vitro BBB models derived from 3xTg-AD astrocytes [78]. Notably, patients with acute glyphosate poisoning show increases in serum s100β [79], which indicates disrupted BBB permeability [80]. Moreover, increased and prolonged neuro- and peripheral inflammation can weaken the BBB [81, 82]. Thus, individuals who exhibit increased inflammation and are vulnerable to disruptions in BBB integrity may be at an increased risk of suffering the deleterious effects due to the ability of glyphosate to infiltrate the brain. A second possibility is that glyphosate crossed through the BBB, as we have demonstrated previously [6], and was subsequently metabolized there. However, this seems less likely as, to our knowledge, there is no evidence that the brain can metabolize glyphosate.

Most glyphosate and aminomethylphosphonic acid studies associate urinary levels of exposure with disease outcomes. No recent studies have investigated bioaccumulation of glyphosate and aminomethylphosphonic acid in tissues as most assume near-complete clearance through standard mammalian excretion routes [83, 84]. Our study is the first to report accumulation of aminomethylphosphonic acid in tissue after a significant recovery period, more specifically in brains of rodents. This finding highlights the need to investigate aminomethylphosphonic acid clearance in tissues and warrants additional studies to determine its molecular interactions and long-term effects on health. Notably, in plants, accumulated aminomethylphosphonic acid has been shown to chelate and sequester bi- and tri-valent metal cations such as Ca²⁺, Zn²⁺, Cu²⁺, Mg²⁺, and Al³⁺ [85]. In humans, heavy metals have been thought to play a role in the pathophysiological process leading to AD [86, 87], suggesting that aminomethylphosphonic acid could chelate select heavy metals in the brain, leading to AD-associated neurotoxicity.

We found long-lasting elevations of various cytokines and chemokines, highlighting neuroinflammation induced by glyphosate exposure in both 3xTg-AD and NonTg mice, consistent with previously-published work showing that glyphosate exposure resulted in increased TNF-α [6]. The increases in several pro- and anti-inflammatory cytokines in the brain suggest that aminomethylphosphonic acid may contribute to dysregulated neuroimmune function months after exposure, although future work is necessary to determine the mechanism of such interaction. Another possibility is that glyphosate exposure may have initiated chronic release of cytokines and chemokines that continues for months. While inflammation is a necessary part of the body’s immune response, prolonged neuroinflammation can be detrimental to the body and brain (reviewed in [88]). Glyphosate has been shown to increase peripheral inflammation: in vitro work demonstrated that exposure of T-cells to glyphosate and its metabolites can directly lead to increased cytokine production [89] and increased reactive oxygen species [90]. Additionally, in mice, short-term administration of glyphosate led to peripheral immune activation and inflammation [91]. 3xTg-AD mice already show heightened peripheral pro-inflammatory responses to infection, with neuroinflammation increasing even when an infection is in the periphery [71]. Given that glyphosate dosing increased peripheral inflammation in our 3xTg-AD mice, despite an absence of circulating glyphosate or aminomethylphosphonic acid, there are three intriguing possibilities to explain the prolonged elevation of peripheral inflammation in the 3xTg-AD mice. Firstly, we did not examine accumulated aminomethylphosphonic acid in other organs; liver and adipose tissue have been shown to cause increased generation of inflammatory cytokines upon glyphosate exposure [92]. Thus, if glyphosate or its metabolites remained in the other organs, it may be continuing to drive chronic peripheral inflammation in the 3xTg-AD mice. Secondly, glyphosate exposure can alter the expression of chromatin remodeling genes in immune cells in vitro [93], which could have led to sustained peripheral inflammation via epigenetic mechanisms. Finally, given the BBB disruptions in the 3xTg-AD strain, it is possible that cytokines generated in the brain from the remaining aminomethylphosphonic acid could have crossed into peripheral circulation, exacerbating inflammatory signaling.

The significant elevations in both pro- and anti-inflammatory cytokines found in the brains of both AD and NonTg mice may have further implications for cognitive dysfunction related to glyphosate exposure. For example, increased levels of plasma Eotaxin and MCP (monocyte chemoattractant protein)-1 are associated with worsening memory in aging adults and those with neurodegenerative diseases, such as AD [94]. We found that both Eotaxin and MCP-1 were significantly elevated in plasma and brain of our glyphosate dosed-mice. Additionally, IFN-γ - a cytokine significantly increased in glyphosate-dosed mice - can induce a neurotoxic phenotype in microglia; during disease states, this can further neuroinflammation and result in oxidative stress, and eventual cell death [95, 96]. Increased IFN-γ expression also impairs adult hippocampal neurogenesis, which can consequently affect learning and memory [96]. Twelve interleukins (IL), a class of cytokines with a wide range of functions centered around the immune response, were significantly increased in cortical tissue of all mice given glyphosate: IL-1⍺, IL-1β, IL-2, -3, -4, -5, -6, -9, -10, -12(p40), -12(p70), -13, and -17. Of these twelve, IL-4, IL-10, and IL-13, are anti-inflammatory cytokines, but prior research has shown that elevated IL-4 and IL-10 can increase Aβ plaque burden [97, 98]. There are mixed conclusions about the implications of individual cytokines on AD; however, the consensus is that the immune response is comprised of a delicate balance of pro- and anti-inflammatory cytokines which is ultimately disrupted in AD and contributes to progression of neuropathology [99, 100]. Since we found that exposure to glyphosate promoted an imbalance of pro- and anti-inflammatory cytokine and chemokine levels in the brain and periphery of 3xTg-AD mice, this may pose profound implications for glyphosate exposure as a risk factor for AD.

Whether through disruption of neuroinflammation, or a more direct mechanism, glyphosate exposure increased Aβ and tau pathology in our 3xTg-AD mice. Notably, amyloidogenic APP processing and tau hyperphosphorylation is worsened with increased cytokine levels [101–103], which was observed in the glyphosate-dosed mice, and we found increased APP processing of C-terminal fragments C99 through elevations in BACE-1. Our group has previously shown that exposing primary neurons from a mouse model of APP overexpression to glyphosate in vitro increased Aβ₄₀ and ₄₂ levels [6]. Additionally, work has shown that N2a cells overexpressing the APP Swedish mutation, then exposed to a combination of IFN-γ plus TNF-⍺, and IFN-γ plus IL-1β, lead to a significant increase in BACE-1 expression and mRNA levels [104]. TNF-⍺, IFN-γ, and IL-1β were all significantly elevated in both blood plasma and cortical tissue from mice exposed to glyphosate, suggesting these increased inflammatory cytokines could be directly affecting the expression of BACE-1. Also, the BACE-1 gene contains a lymphokine response element within its promoter sequence, making it possible for lymphokines to alter the transcription of BACE-1 [104]. Lymphokines include IL-2, -3, -4, -5, and − 6, cytokines which we found to be elevated in mice exposed to glyphosate, offering transcriptional changes as a mechanism for increased BACE-1 expression. We suggest that the accumulation of aminomethylphosphonic acid that remained in the brain during recovery, or glyphosate that may have crossed into the brain during exposure, lead to increased cytokine levels, which increased the expression of BACE-1, increasing amyloidogenic processing and ultimately elevations in amyloidosis. Elevated neuroinflammation may have also contributed to increased tau pathology, as previous work has demonstrated that infection-induced chronic inflammation worsens pathological tau by increasing glycogen synthase kinase-3β activity (GSK-3β), a kinase that can aberrantly contribute to tau hyperphosphorylation [105]. Lastly, exposure to glyphosate and aminomethylphsophonic acid has been linked to transcriptomic changes [6], oxidative damage [54, 75], and disruptions in DNA repair mechanisms [106]. DNA repair mechanisms are critical to safeguard the integrity of a cell’s genetic information. Impaired repair of oxidative DNA damage has been linked to the development and progression of AD [107, 108]. Thus, we speculate that early glyphosate exposure and prolonged aminomethylphosphonic acid in the brain may have contributed to dysregulated DNA damage repair mechanisms, exacerbating AD-like pathology.

We found that glyphosate-exposed 3xTg-AD mice performed similarly in tasks assessing learning and memory compared to the 3xTg-AD 0 mg/kg dosed mice. However, increased thigmotaxia was observed in glyphosate-exposed 3xTg-AD mice. 3xTg-AD mice exhibit significantly increased thigmotaxia in the MWM compared to NonTg mice [109, 110], and aged females, in particular, exhibit a higher level of fear and anxiety demonstrated by increased restlessness, startle responses, and freezing behaviors [111]. A heightened anxiety phenotype is consistent with reports showing that glyphosate exposure increases anxiety-like behaviors in rodents [12, 36, 112]. In humans, amygdalar atrophy and Aβ accumulation are observed early in AD [3, 113, 114], with patients often experiencing anxiety symptoms [114, 115]. 3xTg-AD mice show accumulation of intraneuronal Aβ in the amygdala [116], which has been associated with amygdala-dependent emotional responses and heightened anxiety-like behavior [35, 111, 117]. Additionally, neuroinflammation in the amygdala has been demonstrated to contribute to anxiety-like behavior [118, 119], which in the current study may have extended to the amygdala of 3xTg-AD glyphosate-exposed mice that are already susceptible to anxiety, contributing to increased thigmotaxia in the MWM. Finally, thigmotaxia can be used as a technique when mice cannot locate a hidden platform due to spatial memory impairments [120]. The question remains if the behavioral differences are due specifically to the exposure to glyphosate and accumulated aminomethylphosphonic acid in the brain, the increased AD pathology and neuroinflammation, which may have extended to the amygdala, or a combination of both in 3xTg-AD mice, which will be examined in future work.

Conclusion

In conclusion, glyphosate exposure resulted in premature death, accelerated AD-like pathology and subsequent anxiety-like behaviors in 3xTg-AD mice, and neuroinflammation in both NonTg and AD mice, despite months of recovery (Fig. 6). The multifactorial consequences of glyphosate exposure are increasingly concerning given the ubiquity of its use. The fact that we found accumulation of aminomethylphosphonic acid in brains following recovery from exposure is particularly concerning, especially given its association with increased neuroinflammation in both NonTg and AD mice. As glyphosate use continues to rise, more research is needed to elucidate the impact of this herbicide and its metabolites on the human brain and their potential contribution toward the increased prevalence of neurodegenerative disorders.

Data Availability

The data that supporting the findings of this study will be made available by the corresponding authors upon reasonable request.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AD

Alzheimer’s Disease

US

United States

Aβ

Amyloid Beta

NFT

Neurofibrillary tau tangles

EPA

Environmental Protection Agency

BBB

Blood brain barrier

TNF-⍺

Tumor necrosis factor-⍺

NonTg

Non-transgenic

NOAEL

No observable adverse effect limit

MWM

Morris Water Maze

EDTA

Ethylenediaminetetraacetic acid

MWCO

Molecular weight cut-off

LC-MS/MS

Liquid chromatography/tandem mass spectrometry

RSD

Relative standard deviation

LOD

Lower limit of quantitation

LOQ

Limit of quantitation

ELISA

Enzyme-linked immunosorbent assay

PBS

Phosphate buffered saline

Hp

Hippocampus

Ctx

Cortex

FL-APP

Full-length amyloid precursor protein

GAPDH

Glyceraldehyde 3-phosphate dehydrogenase

BACE-1

Beta-secretase 1

C99

99 amino acid C-terminal fragment of amyloid precursor protein

Aβ₄₀, Aβ₄₂

Amyloid Beta, peptide lengths 40 and 42

Thr 181

Threonine 181

Ser 396

Serine 396

pTau

Phosphorylated tau

UHPLC

Ultra high-pressure liquid chromatography

LC-MRM

Liquid chromatography with multiple reaction monitoring

AT8

Tau phosphorylated at serine 202 and threonine 205

CA1

Cornu ammonis 1

ANOVA

Analysis of variance

UPLC-MS

Ultra-performance liquid chromatography- mass spectrometry

APP

Amyloid Precursor Protein

G-CSF

Granulocyte-colony stimulating factor

IFN-γ

Interferon-gamma

MHC

Major histocompatibility complex

IL

Interleukin

KC

Keratinocyte-derived cytokine

MCI

Mild cognitive impairment

KC/CXCL1

Chemokine ligand 1

MIP1-β

Macrophage inflammatory protein 1-β

MCP-1

Monocyte chemoattractant protein 1

Author contributions

SKB: Animal dosing, behavioral testing, tissue processing, wrote and edited manuscript. WW: Biochemical analysis, statistical analysis, wrote and edited manuscript. RS: Assistance with UPLC measurements, edited the manuscript. KVP: UPLC measurements, edited the manuscript. ST: Behavior testing, statistical analysis, wrote and edited the manuscript. JJ: Behavior testing, histology, stereology, wrote and edited the manuscript. HL: Histology, edited the manuscript. JT: Histology, western blots, edited the manuscript. PP: Funding, experimental design, statistical analysis, wrote and edited the manuscript. RV: Funding, experimental design, statistical analysis, wrote and edited the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by grants to Ramon Velazquez from the ASU Edson Initiative Seed grant program and the National Institute on Aging (R01 AG059627) and (R01 AG062500). This reported research includes work performed in the Integrated Mass Spectrometry Shared Resource supported by the National Cancer Institute of the National Institutes of Health under grant number P30CA033572.

Data availability

No datasets were generated or analysed during the current study.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.