High dietary phosphate intake induces anxiety in normal male mice

Abstract

Background

Diet is increasingly recognized as an important risk factor for mental health. Inorganic phosphate (Pi) is currently used as a flavor enhancer or preservative at an unregulated amount in the western diet despite evidence that excessive dietary Pi intake associates with metabolic and cardiovascular disorders. The impact of high Pi on brain function remains poorly understood. This study aimed to evaluate the effects of chronic consumption of high dietary phosphate on behavior, neurovascular health, and neuroimmune populations, and cortical gene expression in key brain regions associated with emotional regulation.

Methods

Adult C57BL/6 male mice were fed either a normal phosphate (NP) or high phosphate (HP) diet for 12 weeks. Behavioral assessments included the open field test (OFT) and fear conditioning. Histological analyses assessed neuronal densities and vascularization. Flow cytometry quantified brain-resident immune cell populations and microglia. Unbiased analysis of hippocampal gene expression was performed using RNA sequencing (RNA-Seq).

Results

HP-fed mice exhibited increased anxiety-like behaviors compared to NP-fed controls, as indicated by increased thigmotaxis (i.e., more time close to the walls and, consequently, less time spent in the central area, HP: 164 ± 61 vs. NP: 215 ± 54 s, P = 0.03) in the OFT and increased time freezing regardless of stimulus type during fear conditioning. Neuronal density is significantly decreased in the hypothalamus of HP-fed mice (21.9 % ± 4.5 % vs. 16.4 ± 2.9 %, P = 0.02) but without concomitant differences in brain vascularization. Immunophenotyping showed that HP-diet significantly reduced TCRβ⁺ T cells and NK1.1⁺ NK cells (both P < 0.05), suggesting diet-induced alterations in neuroimmune homeostasis. RNA-Seq identified significant alterations in gene expression in the hippocampus, including upregulation of Neat1 and Stat3 and downregulation of Igf2, which are implicated in stress regulation, neurodegeneration, synaptic plasticity and immune system pathways.

Conclusions

Collectively, this study highlights that habitual consumption of high dietary phosphate in mice may induce chronic anxiety, accompanied by significant changes in the neuronal and brain-resident immune populations. The data point to a potential link between dietary Pi, neuroinflammation, and the pathogenesis of anxiety and depression in otherwise healthy young male mice. Given the prevalence of phosphate additives in processed foods, these findings have important public health implications supporting the regulation of Pi in the food industry.

Highlights

• •Chronic high dietary phosphate consumption induced anxiety-like behaviors in mice.
• •Abnormal behaviors were accompanied by altered brain neuronal density.
• •Brain homeostatic immune cell composition was affected by excess phosphate intake.
• •High phosphate intake altered hippocampal gene expression linked to stress regulation.

1. Introduction

The World Health Organization estimates that over 970 million people worldwide suffer from a mental health disorder, with depression and anxiety being major contributors to disease burden and disability (Global and regional, 2022). Although the pathophysiology of anxiety and depression is multifactorial in nature, an increasing number of studies have linked unhealthy dietary patterns to the development of mental disorders (Sánchez-Villegas et al., 2011; Fatahi et al., 2021). Specifically, epidemiological studies highlight a concerning association between the consumption of ultra-processed foods, often rich in inorganic phosphate (Pi), and the development of anxiety and depression in the general population (Hecht et al., 2022; Sun et al., 2023).

Pi is an essential biological molecule involved in energy metabolism, cell signaling, and bone mineralization (Jacquillet and Unwin, 2019). However, its excessive use as a preservative and flavor enhancer in processed foods has led to a significant increase in dietary Pi intake. It is estimated that phosphate additives are present in 56 % of the top 25 US food and beverage manufacturers in 2020 (Dunford and Calvo, 2025). Currently, the recommended dietary allowance (RDA) for phosphate is 700 mg while the National Health and Nutrition Examination Survey found that the average phosphorus intake in US adults is nearly twice that, at 1368 mg of phosphorus/d (McClure et al., 2017). Furthermore, estimates suggest that up to 25 % of all US adults consume Pi at levels 3–5 times higher than the recommended daily allowance (Sullivan et al., 2023; Lee and Cho, 2015). While the detrimental effects of a high-phosphate (HP) diet are already documented in the context of aging, cardiovascular health, and kidney function (Mizuno et al., 2016; Hu et al., 2015), the long-term impact on brain function and mental health remain largely unexplored.

Evidence from in vitro and animal studies indicates that excessive Pi intake can impair angiogenesis, reduce vascular density, and promote cell death and inflammation in the peripheral vasculature (Erem et al., 2022; Román-García et al., 2010). Studies in the human brain microvascular endothelial cells indicated that high phosphate milieu downregulates expression of tight junction proteins, including zona occludens-1, occludin, and claudin-5, which play a major role in the blood-brain barrier (BBB) integrity (Chung et al., 2019). These neuroinflammatory and structural alterations are recently recognized in the pathogenesis of mood disorders, including anxiety and depression (Guo et al., 2023; Won and Kim, 2020). Furthermore, consumption of fast food and processed foods has been linked to anxiety symptoms in young adults (Tang and Lee, 2024; Xie et al., 2024). However, direct evidence linking HP-diet to anxiety-like behaviors and neurobiological changes in brain regions associated with mood regulation remains lacking.

To address this gap, we conducted an experimental study to assess the direct impact of chronic HP-diet consumption on anxiety-related behaviors and neurobiological alterations in mice. We investigated changes in behavior, neuroimmune interactions, and gene expression in brain regions critically involved in mood regulation, including the hippocampus, hypothalamus, and amygdala. Our findings reveal that prolonged high-phosphate intake induces anxiety-like behaviors, alters gene expression patterns in the hippocampus, and modifies immune cell populations, contributing to a pro-inflammatory state in the brain in otherwise healthy mice.

2. Materials and methods

2.1. Animal experiments

All animal procedures were approved by and conducted under the oversight of the UT Southwestern Institutional Animal Care and Use Committee (IACUC) following NIH guidelines for the care and use of laboratory animals. Experiments were performed using 20- to 24-week-old male C57BL/6J mice obtained from the UT Southwestern breeding core. Mice were housed in a temperature- and humidity-controlled facility on a 12-h light/dark cycle with ad libitum access to food and water. Mice were randomly assigned to either a normal phosphate (NP, n = 27) or HP (n = 24) diet for 12 weeks. The HP-diet (TD.08020, Envigo Teklad, Madison, WI) contained 2.0 % inorganic phosphate (total phosphate: 2.3 %), which is equivalent to ∼3-5 × the FDA-recommended maximum human daily intake. The NP diet (TD.160114) contained 0.6 % inorganic phosphate (total phosphate: 0.9 %). Both diets were matched for other mineral contents (0.3 % magnesium, 1.9 % calcium, 1.8 % potassium, and 0.9 % sodium). Mice were weighed weekly throughout the 12-week dietary intervention period.

2.2. Behavioral testing

All behavioral tests were performed during the light phase (between 09:00–17:00) under consistent conditions. Mice were habituated to the testing room for at least 30 min before each test.

Open field test: The open field test (OFT) assessed locomotor activity and anxiety-like behavior. Mice fed a NP (n = 15) diet or a HP (n = 12) diet were placed in the periphery of a novel open field environment (44 cm × 44 cm, walls 30 cm high) in a dimly lit room (60 lux) and allowed to explore for 10 min. The animals were monitored from above by a video camera connected to a computer running video tracking software (Ethovision XT 13.0 Noldus, Leesburg, VA) to determine the time and distance moved in two areas: the periphery (5 cm from the walls) and central or non-periphery (34 cm × 34 cm). Time spent in the periphery and non-periphery as well as total distance moved in NP vs. HP group were compared. The arena was cleaned with disinfectant and allowed to dry between mice.

Elevated Plus Maze test: The elevated plus maze (EPM) assessed anxiety-like behavior. Mice fed a NP (n = 15) diet or a HP (n = 12) diet were placed at the center of a plus maze with two dark enclosed arms and two open arms elevated 50 cm above the floor in a dimly lit room and allowed to explore freely for 5 min. The animals were monitored from above by a video camera connected to a computer running video tracking software (Ethovision XT 13.0 Noldus, Leesburg, VA). Time spent in the open and closed arms as well as in the middle were recorded. The apparatus was wiped with disinfectant and allowed to dry between mice.

Fear Conditioning: Fear conditioning was measured in boxes equipped with a metal grid floor connected to a scrambled shock generator and freezing was measured by the VideoFreeze software (Med Associates Inc., St. Albans, VT) in 23 NP animals or 21 HP animals. On the first day (training), mice were individually placed in a test chamber. Baseline anxiety levels were assessed and calculated as the percentage of time spent freezing during the first 2 min (i.e., habituation phase). After 2 min, the mice received 3 tone-shock pairings (30 s white noise, 80 dB tone co-terminated with a 2 s, 0.5 mA footshock, 1 min intertrial interval). On the second day (contextual fear conditioning test), memory of the context was measured by placing the mice into the same chambers and freezing was measured for 5 min. Forty-eight hours after, memory for the white noise cue was measured by placing the mice in the test chamber with altered floors and walls, and a vanilla smell (cued fear conditioning test). Freezing was measured for 3 min, then the noise cue was turned on for an additional 3 min and freezing was measured. The percentage of time spent freezing is calculated as the time freezing (lack of all movement except respiration) relative to the total time of each test segment (5 min for the contextual test and 3 min each for the pre-tone and during tone of the cued test).

Pain sensitivity (pain threshold): Pain sensitivity was assessed using an incremental foot shock paradigm in 15 NP mice or 12 HP mice. The mice were placed individually into boxes equipped with a metal grid floor connected to a scrambled shock generator (Med Associates Inc., St. Albans, VT). After ∼1 min, the mice received a series of foot shocks (2 s each) with increasing intensity. The initial shock intensity was 0.05 mA, and the amplitude was increased by 0.05 mA for each consecutive foot shock with 15 s inter-shock intervals. The lowest shock intensity that each animal displayed each behavior (flinch, jump, and vocalization) was recorded.

2.3. Immunohistological (IHC) analysis

Brain hemispheres collected from 10 NP mice and 10 HP mice were post-fixed in 4 % PFA, and cryoprotected in 30 % sucrose in 0.1 M PBS. The bisected brains were sectioned serially and coronally, using a Leica CM1860 cryostat (Leica Biosystems GmbH, Nussloch, Germany) at a thickness of 30 μm, and were immersed into a 12-well plate containing cryoprotectant solution and kept in a −20 °C freezer, until needed. The sections were washed in 1X PBS and mounted onto glass microscope slides. They were air-dried and placed briefly (1–2 min at 37 °C) inside a gravity convection oven to ensure they were fully dry. Each animal had sections stained with both alkaline phosphatase and cresyl-violet; alkaline phosphatase was used to stain the capillary networks of the brains, and cresyl-violet their neurons. For Alkaline Phosphatase (Capillary Density), slides were incubated with staining substrate solution for 2 h, rinsed in 1X PBS, and dehydrated in intervals with varying concentrations of ethanol before cover slipping with Permount. For Cresyl-violet (Cell Density), sections were stained with 10X cresyl-violet for 5 min, rinsed, dehydrated, and moved to xylene for 3 min before mounting with Permount.

Area fraction analysis: Brain sections were scanned at 20 × magnification using an Axio Scan.Z1 Slide Scanner (Zeiss Microscopy GmbH, Germany) to generate high-resolution images for quantitative histological analysis. The final z-stack generated was achieved at constant 2 μm step size with a total of 14–15 optical slices. Area fraction analysis was performed for both alkaline phosphatase (vascular staining) and cresyl-violet (neuronal staining) using HALO v2.3 (Indica Labs Inc. Albuquerque, NM, USA), by an observer blinded to experimental groups. The assessment of cresyl-violet stained cells was performed on maximum intensity z-stack projections. A region of interest (ROI) corresponding to the cortex, corpus callosum, striatum, hippocampus, hypothalamus and amygdala was drawn based on the Allen Mouse Brain Atlas (2015). Each scanned image was annotated to delineate distinct brain regions of interest. Separate annotation layers were generated for region-specific analysis of neuronal and capillary density while eliminating confounding factors such as tissue damage, sectioning artifacts, air bubbles, or blurred areas that could distort automated image analysis. The real-time tuning feature within HALO was employed to optimize optical density thresholds, allowing for the selective detection of positively stained areas while minimizing background noise and artifacts. Each annotated layer was then analyzed separately, and the HALO algorithm computed the percentage of positively stained area, providing an objective measure of neuronal or vascular density. To ensure consistency and eliminate observer bias, automated segmentation parameters were set uniformly across all images, and final results were reviewed for accuracy. The classification of cells as neurons was contingent upon the presence of a nucleolus, dendritic processes, euchromatin material within the nucleus, and nuclei enveloped by cytoplasm (García-Cabezas et al., 2016). The percentage of positive stained area was used to represent the density of the neurons or capillaries.

2.4. Fluorescence-activated cell sorting (FACS) analysis

Following final behavioral testing, mice were anesthetized with isoflurane and perfused transcardially with cold 0.1 M PBS (∼30 ml). Brains were divided by hemisphere, with one hemisphere (left) going toward FACS analysis (Malone et al., 2023). FACS analysis was performed on hemi-sectioned brains from 13 NP animals and 11 HP animals. Brain samples were homogenized, centrifuged, and filtered through 70 μm filters. The samples then went through a 30 % Percoll gradient to isolate leukocyte cells by size and remove myelin and debris. Cells were washed in 1X PBS and incubated in Ghost Dye 780 (Tonbo Biosciences) with PBS in the dark at 4 °C for 30 min. Cells were washed twice in FACS buffer (1X PBS, 1 % Bovine serum albumin, 0.01 % sodium azide). Afterwards, cells were blocked using a FcR blocking reagent for 5 min at room temperature. Antibodies were added without washing and incubated in the dark at 4 °C for 30 min. The cells were stained using a general immunophenotyping panel targeting CD45, TCRβ, CD4, CD8, CD19, CD11b, NK1.1, Ly6C/Ly6G, and CD138. Cells were washed twice in FACS buffer and fixed in 1 % paraformaldehyde plus 0.1 % EDTA at 4 °C for 30 min. Sample acquisition was conducted using a FACSymphony (BD Biosciences) flow cytometer and BD FACS DIVA software. All gating and analyses were performed in FlowJo v10. Cell populations were identified using expression profiles through manual gating (Fig. S1).

2.5. RNA sequencing, gene omnibus, and pathway analysis

Hippocampi from mice fed a NP (n = 4) diet and a HP (n = 3) diet were collected for RNA sequencing. RNA was extracted and purified with RNA Direct-zol kits (Zymo Research, US; R2050) as we described previously. (Wyler et al., 2024) (Pertea et al., 2016)^, (Kim et al., 2019) The gene expression levels based on fragments per kilobase of transcript per million read pairs (FPKM) were calculated using StringTie v2.2.1 (Pertea et al., 2015). Differential expression genes were extracted using the Ballgown package v2.28.0 in R (v3.6.0). A heatmap and a volcano plot were generated using R. Differential expression was determined based on fold change (FC) between HP and NP groups (HP/NP). Genes with |log2 FC| > 0.58 and p < 0.05 were considered significantly differentially expressed. These genes were classified into two groups: upregulated (log2 FC > 0.58 with p < 0.05) and downregulated (log2 FC < −0.58 with p < 0.05), and Gene Ontology (GO) analysis was performed separately for each group using DAVID (v2022q4) (Sherman et al., 2022). GO terms with p < 0.05 were considered significantly enriched.

2.6. Statistical analysis

Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) to evaluate differences in behavioral responses between the NP- and HP-diet groups. A linear mixed model was applied to assess changes in behavioral outcomes, incorporating treatment group (NP vs. HP), behavioral test type, and group-by-test interaction as fixed effects. If model assumptions were not met, log transformation was applied to normalize the data. Statistical significance was defined as p < 0.05. To analyze neuronal and capillary densities, and WB results, GraphPad Prism version 10.2 (GraphPad, San Diego, CA, USA) was used to compare the average percentage of positive stained areas between groups. Group comparisons were conducted using an unpaired t-test with Welch's correction, except for alkaline phosphatase-stained sections of the cortex and striatum, where a Mann-Whitney test was employed due to non-normal data distribution.

3. Results

3.1. HP-diet induces general anxiety-like behaviors in otherwise healthy mice

In the OFT assessing anxiety-like behavior, HP-fed mice spent significantly less time in the non-peripheral (i.e. central) area compared to NP-fed mice (164 ± 61 vs. 215 ± 54 s, P = 0.03, Fig. 1A). Importantly, the total distance moved in the open field was not significantly different between the 2 groups (HP: 3662 ± 708 cm vs. NP: 4182 ± 938 cm, Fig. 1A), suggesting that general locomotor activity was unaffected by dietary phosphate intake. In the EPM test, a second test that measures generalized anxiety, there were no significant differences between the HP and NP groups in the time spent in the open arms (Fig. 1B), closed arms, or center (Fig. S2A) as well as in the total distance moved (Fig. 1B), indicating that long-term HP consumption did not alter exploratory behavior in this task.

Fig. 1.

Anxiety-like behavior assessments in normal phosphate (NP)- and high phosphate (HP)- fed mice. A) Graphs showing cumulative time spent in the non-periphery and total distance moved by mice fed a NP diet (n = 15) or a HP diet (n = 12) in the open field test. B) Graphs showing cumulative time spent in the open arms and total distance moved by mice treated with NP (n = 15) or HP (n = 11) in the elevated plus maze test. C) Quantification of the percentage of time spent freezing during the cued fear conditioning test in mice fed a NP diet (n = 23) or a HP diet (n = 21). D) Graph showing foot shock sensitivity, tested using a flinch, vocalize, jump procedure in NP (n = 15) and HP (n = 12) mice, and mice received a series of foot shocks with increasing intensity in 0.05 mA increments. ∗P < 0.05 vs. NP.

We also tested the mice using fear conditioning, a classical aversive test that measures learning and memory and which we have previously shown to be altered by changes in the adaptive immune system after ischemic stroke (Ortega et al., 2020). During the cued fear conditioning test (on the third day), mice in the HP group froze significantly more compared to the NP group (before tone: 29.67 ± 25.87 s vs. 17.64 ± 11.46 s; during tone: 104.1 ± 27.40 s vs. 92.08 ± 26.83 s, group effect: P = 0.03, Fig. S2B). The percentage of time spent freezing was also significantly higher in the HP group compared to the NP group during the cued fear conditioning test (before tone: 16.5 ± 14.4 % vs. 9.80 ± 6.37 %; during tone: 57.8 ± 15.2 % vs. 51.2 ± 14.9 %, group effect P = 0.03, Fig. 1C). Similarly, freezing duration and the percentage of time spent freezing tended to be higher in the HP group than the NP group during the contextual fear conditioning test (on the second day), although the difference did not reach statistical significance (freezing duration: 119.0 ± 51.37 s vs. 92.93 ± 36.95 s, P = 0.058; percentage of time freezing: 39.7 ± 17.1 % vs. 31.0 ± 12.3 %, P = 0.058, Fig. S2C). There were no significant differences in freezing duration and the percentage of time spent freezing during the 2-min habituation phase on the first day (i.e., training phase) between NP and HP mice (Fig. S2D). These findings suggest an enhanced fear response in HP-fed mice. However, this difference was not attributed to increased pain sensitivity, as the minimal foot shock amplitudes required to induce flinching, vocalization, and jumping were not significantly different between the two groups (Fig. 1D).

3.2. High dietary Pi consumption alters neuronal densities in discreet brain regions

To determine the influence of high dietary Pi intake on the brain structure that may underlie behavioral phenotype, we first quantified neuronal density using cresyl-violet staining in whole-brain hemispheric sections (Fig. 2). Cresyl-violet selectively stains Nissl substance, a marker for neuronal populations, allowing for the visualization and quantification of regional cell density differences (Zhang et al., 2021). Analysis revealed that neuronal densities in the cortex, corpus callosum, and hippocampus were not significantly altered by dietary phosphate levels (Fig. 2). However, in the hypothalamus, HP-fed mice exhibited ∼25 % reduction in neuronal density compared to NP-fed controls (21.9 % ± 4.5 % vs. 16.4 ± 2.9 %; P = 0.02).

Fig. 2.

A representative image of a cresyl-violet-stained brain section used for neuronal density analysis with annotated hemispheric regions of interest (ROIs) (A). After optical density threshold optimization for background noise removal, positively segmented neurons (red-yellow) were quantified in the hippocampus (B), hypothalamus (C), corpus callosum, cortex, striatum, and amygdala. Representative images of alkaline-phosphatase-stained brain sections used for vascular density analysis showing positively segmented vessels (red-yellow) in the cortex (D) and corpus callosum (E). Percentage of cell-populated area across different ROIs (F) in mice fed a normal phosphate (NP) diet or a high phosphate (HP) diet, including cortex (NP, n = 9 and HP, n = 10), striatum (NP, n = 9 and HP, n = 9), corpus callosum (NP, n = 10 and HP, n = 10), hypothalamus (NP, n = 7 and HP, n = 7), hippocampus (NP, n = 6 and HP, n = 7), and amygdala (NP, n = 8 and HP, n = 10). Neuronal density in the hypothalamus was significantly decreased, while small but significant increases were observed in the striatum and amygdala of HP-fed mice compared to NP-fed mice. ∗P < 0.05 vs. NP. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

In contrast, neuronal density was modestly, but significantly, increased in the striatum (Welch's test; F(8,8) = 3.368; P = 0.02) and amygdala (Welch's test; F(9,6) = 3.766; P = 0.03) in HP-fed mice compared to NP-fed mice. These findings suggest that phosphate intake may influence neuronal proliferation, survival, and/or structural remodeling in regions associated with motor function (striatum) and emotional processing (amygdala).

3.3. High Pi consumption does not induce significant changes in overall brain vascularization

To determine whether HP intake affects brain vascularization that may subsequently affect neuronal density, we analyzed capillary density using alkaline-phosphatase staining, a widely used histochemical marker for cerebral microvasculature. Image analysis of stained tissue (e.g., cortex and corpus callosum, Fig. S3), showed no significant differences in capillary density between NP- and HP-fed mice across all examined brain regions, including the cortex, corpus callosum, hippocampus, and striatum (P > 0.99). These findings suggest that chronic high phosphate intake does not lead to global or regional alterations in cerebrovascular microvessel density.

3.4. Effects of dietary Pi consumption on brain immune cell populations

Given the established relationship between CD4⁺ T cells and stress-induced anxiety-like behaviors(Fan et al., 2019), we sought to determine whether prolonged HP intake alters the homeostatic immune cell composition within the brain. To address this, we analyzed leukocyte subpopulations in normal mice maintained on either HP or NP diets for three months, using flow cytometry-based immunophenotyping. The immune cell populations assessed included CD45⁺TCRβ⁺CD4⁺ (CD4⁺ T cells), CD45⁺TCRβ⁺CD8⁺ (CD8⁺ T cells), CD45⁺CD19⁺ (B cells), CD45⁺CD19⁻ TCRβ⁻ NK1.1⁺ (Natural Killer [NK] cells), CD45⁺CD11b⁺Ly6C⁺ (Neutrophils/Eosinophils), and CD45⁺CD11b⁺Ly6G⁺ (Monocytes/Macrophages). Analysis of the percentage of total immune cells revealed a significant reduction in total T cell population representation in HP-fed mice compared to NP-fed controls (P = 0.02, Fig. 3). However, when examining CD4⁺ and CD8⁺ subpopulations separately, no significant changes were observed between dietary groups. This suggests that the overall reduction in T cells may be driven by a decrease in other non-canonical T-cell subsets rather than CD4⁺ or CD8⁺ cells, specifically. In addition, the percentage of natural killer (NK) cells was significantly lower in the brains of HP-fed mice compared to NP-fed mice (P = 0.03), suggesting that dietary phosphate intake may influence the innate immune response within the brain. CD19⁺ B cell and neutrophil populations exhibited an increasing trend in HP-fed mice, though these differences did not reach statistical significance.

Fig. 3.

Brain immune cell populations in normal phosphate (NP)- and high phosphate (HP)-fed mice after three months of dietary exposure. Flow cytometry analysis was performed to assess the proportions of various leukocyte subsets in brain tissue including CD45⁺ leukocytes (NP, n = 13 and HP, n = 11), B cells (NP, n = 13 and HP, n = 11), T cells (NP, n = 13 and HP, n = 11), CD4⁺ T cells (NP, n = 13 and HP, n = 11), CD8⁺ T cells (NP, n = 13 and HP, n = 11), NK cells (NP, n = 13 and HP, n = 10), neutrophils (NP, n = 13 and HP, n = 11), monocytes/macrophages (NP, n = 13 and HP, n = 11), and microglia (NP, n = 13 and HP, n = 11). ∗P < 0.05 vs. NP.

3.5. Dietary Pi alters RNA expression in hippocampus

Given the significant differences in behavioral patterns suggestive of increased anxiety-like behavior in HP-fed mice, we conducted an unbiased transcriptome-wide RNA sequencing (RNA-Seq) to identify potential molecular mechanisms underlying these behavioral changes in the hippocampus, a brain region involved in anxiety-related behavior and stress responses. RNA-Seq analysis revealed 30 differentially expressed gene (DEGs) that met the cutoff criteria (|log2 FC| > 0.58 with P < 0.05) were identified (12 upregulated, 18 downregulated, Fig. 4A–B and Table S1) in the hippocampus (n = 4 for HP and 3 for NP). Among the significantly altered genes, nuclear-enriched abundant transcript 1 (Neat1), MCTP1, and Stat3 were upregulated in the hippocampus of HP-fed mice. Conversely, insulin-like growth factor 2 (Igf2) was downregulated (Fig. 4B and Table S1). GO analysis revealed five significantly enriched GO terms associated with 12 upregulated genes and four with 18 downregulated genes (Fig. 4C).

Fig. 4.

Transcriptomic profiling of hippocampus from normal phosphate (NP, n = 4)- and high phosphate (HP, n = 3)-fed mice. A) Heat maps showing a comparison of normalized differential gene expression values of the 30 differentially expressed enriched genes (DEGs) between the NP and HP groups. B) Volcano plot showing log2 fold change (FC) values for all genes with highlighting for those that are significantly upregulated (red dots) or downregulated (blue dots). The criteria for upregulated genes are log2 FC > 0.58 with P < 0.05, and for downregulated genes, log2 FC < −0.58 with P < 0.05. C) Gene Ontology analysis of DEGs that are upregulated or downregulated in the hippocampus with the HP diet. The criteria for upregulated and downregulated genes are the same as in B. Biological process, cellular component, and molecular function are represented in red, blue, and green, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

4. Discussion

The major findings from our study are four-fold: First, we demonstrate a direct effect of dietary Pi overconsumption on development of anxiety in otherwise normal mice. Second, the behavioral change is accompanied by changes in the neuronal density in the multiple brain regions involved in mood regulation, including amygdala and striatum, which cannot be attributed to changes in cerebral microvasculature. Third, HP diet induces changes in brain-resident immune populations, including reductions in T cells and NK cells, which may contribute to the behavioral phenotype. Fourth, the changes in neuronal and immune cell composition are accompanied by dysregulation of genes involved in neuronal cell migration, excitability, and vesicle traffic function in the hippocampus, the major brain region involved in mood regulation.

The observed behavioral changes in HP-fed mice, including increased peripheral and reduced central activity in the OFT, suggest increased thigmotaxis (a preference for the edge of the field), which is a hallmark of anxiety-like behavior (Simon et al., 1994). Although significantly increased freezing observed during the white noise tone of the fear conditioning test may signify heightened fear learning in the HP group, higher rates of freezing were also observed in the HP mice prior to the tone,without significant group by tone interaction. Thus, the generalized increase in freezing regardless of stimulus (before and after the tone during cued fear conditioning test) provides further evidence for increased anxiety rather than enhanced learning or memory in the HP group. Increased nociception could theoretically contribute to this behavior, but this explanation is unlikely since pain sensitivity thresholds were not different between HP and NP groups. On the other hand, there was no group difference in the time spent in the open arm of elevated plus maze, suggesting that the HP diet-induced anxiety is moderate and potentially context-dependent. Previous studies have indicated that behavioral response to OFT test is mainly dependent on the tactile cue from the vibrissae that can be reduced when whiskers are removed (Seibenhener and Wooten, 2015). Conversely, the EPM responses reflect risk taking behavior, which are influenced by multiple sensory and cognitive pathways beyond vibrissae inputs. These differences may explain the lack of significant findings in this test (Filgueiras et al., 2014).

The mechanisms underlying HP diet-induced anxiety remain unclear but may involve alterations in neuronal integrity, neuroinflammation, and synaptic remodeling. The observed reduction in neuronal density in the hypothalamus suggests that chronic phosphate intake may disrupt neuroendocrine regulation via the hypothalamic-pituitary-adrenal (HPA) axis, a key system controlling stress responses (Smith and Vale, 2006). Consistent with this, metabolic stressors have been shown to impair hypothalamic neurogenesis and gliogenesis, leading to dysregulation of homeostatic processes governing behavior and endocrine function (Recabal et al., 2017). Conversely, the increase in neuronal density within the amygdala is consistent with chronic stress-induced neuroanatomical remodeling, where hyperactivity is often associated with anxiety disorders (Vyas et al., 2004). Chronic stress exposure has been shown to drive dendritic hypertrophy and increased neuronal activity in the amygdala, but not in the hippocampus, which may contribute to heightened anxiety responses (Peay et al., 2023). The observed increase in amygdala neuronal density, coupled with behavioral alterations, suggests that HP diet may potentiate fear-related neural circuits, increasing threat sensitivity and stress reactivity in otherwise healthy young mice.

Mechanisms underlying altered regional neuronal density induced by HP diet is unknown. Excessive dietary phosphate has been linked to endothelial dysfunction(Mizuno et al., 2016; Uribarri and Calvo, 2013; Ohnishi and Razzaque, 2010), which may influence brain microvasculature. However, the capillary density in multiple brain regions was unaffected by the HP intake. The absence of significant vascular changes suggests that HP diet may not induce overt microvascular rarefaction. Alternatively, neuroinflammation may lead to neuronal damage and synaptic dysfunction, particularly in brain regions implicated in anxiety regulation, such as the hippocampus, hypothalamus, and amygdala (Won and Kim, 2020; Zheng et al., 2021; Miller and Spencer, 2014; Cai, 2013). High dietary Pi intake has been shown to alter gut microbiota in both healthy rats and rats with chronic kidney disease (Ye et al., 2021), which may contribute to altered neuronal density or plasticity in the brain regions involved in mood regulation via gut brain axis communication (Bertollo et al., 2025). Previous studies have demonstrated that CD4⁺ T cells, play a crucial role in modulating stress-induced anxiety-like behaviors (Fan et al., 2019). Regulatory T cells (Tregs), in particular, serve a protective role by suppressing excessive neuroinflammation, and their depletion has been linked to heightened susceptibility to anxiety (Kim et al., 2012). Our findings indicate that HP-diet significantly decreased total T-cell representation, though specific CD4⁺ and CD8⁺ subpopulations did not show consistent changes. This overall reduction in T cells may suggest a shift in pro-/anti-inflammatory homeostasis, potentially altering neuroimmune interactions relevant to anxiety disorders. Furthermore, NK1.1 (CD161⁺) natural killer (NK) cells were significantly reduced in the brains of HP-fed mice. Prior research has shown that NK cell depletion correlates with higher depressive symptom scores in patients with major depressive disorder(Savitz et al., 2013), suggesting that NK cell function may be linked to emotional resilience. However, the direct contribution of NK cell loss to anxiety-like behaviors in our study remains unclear.

The differential gene expression in the hippocampus induced by HP intake provides key insights into molecular mechanisms linking dietary phosphate intake to anxiety. Upregulation of Neat1, a lncRNA involved in neuronal excitability and calcium homeostasis in the hippocampus may be responsible for anxiety response observed in HP-fed mice (Kukharsky et al., 2020a). This is supported by studies demonstrating that Neat1 knockout (Neat1^−/−) mice exhibit reduced anxiety-like behavior in the EPM test (Kukharsky et al., 2020b). MCTP1, a neuronal vesicle/endosome protein expressed in hippocampus and amygdala, is also upregulated in the mice treated with HP diet. Overexpression of MCTP-1 was shown to impair neuronal cell migration and synaptic vesicle formation and recycling, which may vulnerability to neuropsychiatric diseases (Qiu et al., 2015). Additionally, the upregulation of Stat3, which is known to play a role in neuroinflammation, suggests that HP intake may impair neuroinflammatory responses involved in anxiety-modulation. Stat3 knockdown mice, achieved by AAV-Stat3 shRNA, exhibit alleviated lipopolysaccharide-induced anxiety-like behavior in the OFT and EPM along with inhibited inflammasome activation in the hippocampus (Shentu et al., 2024). The downregulation of Igf2, a peptide hormone involved in glucose and lipid metabolism which also plays a critical role in the synaptic transmission and consolidation of fear memories(Chen et al., 2011), further supports the notion that HP intake negatively impacts cognitive and emotional resilience. Collectively, our results reveal that the HP diet leads to dysregulation of gene expression pathways involved in the regulation of neuronal excitability, cellular trafficking, and synaptic function, which may be responsible for increased fear learning and retention.

Several limitations in our study should be considered. We primarily studied that 20- to 24-week-old male mice, and the results may not be applicable to younger or older mice as well as female mice. While we detected changes in neuronal density in key brain regions, the underlying mechanisms, whether driven by altered neurogenesis, apoptosis, or synaptic remodeling, remain unclear and require further investigation using cell-specific markers. Additionally, our RNA-Seq analysis was limited to the hippocampus, leaving potential gene expression changes in other brain regions (e.g. hypothalamus, amygdala) unexplored. Finally, although we observed immune cell alterations, their functional impact remains uncertain; future studies should assess cytokine profiles, BBB integrity, neuroimmune interactions, and gut microbiota to clarify the link between phosphate intake and neuroinflammation.

5. Conclusions

Our study provides the first direct evidence of the detrimental effect of HP diet in the development of anxiety. Given the widespread use of phosphate additives in processed foods and the lack of mandatory labeling of the Pi content in the food supply, our study findings may have important public health implications. The high bioavailability and rapid absorption of inorganic phosphate may fuel the epidemic of mood disorders in modern societies that heavily rely on processed food consumption. Additional studies in humans are needed to confirm the causal role of dietary phosphate excess in the development of anxiety.

CRediT authorship contribution statement

Pavel Yanev: Writing – review & editing, Visualization, Investigation, Formal analysis. Thomas A. Ujas: Writing – review & editing, Writing – original draft, Visualization, Investigation, Formal analysis. Han-Kyul Kim: Writing – review & editing, Writing – original draft, Visualization, Investigation, Formal analysis. Teppei Fujikawa: Writing – review & editing, Visualization, Investigation. Noriyoshi Isozumi: Writing – review & editing, Visualization, Investigation. Eiichiro Mori: Writing – review & editing, Supervision. Jadwiga Turchan-Cholewo: Writing – review & editing, Investigation. Connor Stuart: Writing – review & editing, Investigation. Rowan Sturgill: Investigation, Writing – review & editing. Shari G. Birnbaum: Writing – review & editing, Investigation. Ann M. Stowe: Writing – review & editing, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Funding acquisition, Formal analysis. Wanpen Vongpatanasin: Writing – review & editing, Writing – original draft, Visualization, Validation, Resources, Project administration, Methodology, Funding acquisition, Formal analysis, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the supplementary data to this article:

Data availability

Data will be made available on request.