A novel Mediterranean diet-inspired supplement reduces hippocampal amyloid deposits and microglial activation through the modulation of the microbiota gut-brain axis in 5xFAD mice

5xFAD mice with a C57Bl/6 background overexpress human amyloid precursor protein (APP) with three FAD mutations [the Swedish (K670N, M671L), Florida (I716V), and London (V7171) mutations] and human PSEN1 with two FAD mutations (M146L and L286V).27 The 5xFAD mouse model of AD displays intraneuronal amyloid beta (Aβ) from 6 weeks of age, neuronal loss in the subiculum and cortical layer V at approximately 39 weeks24 and hippocampal-dependent cognitive deficits from 26 to 56 weeks of age.20^,21 16 males and 16 females 5xFAD mice (6 weeks old) sourced from Jackson Laboratories (Bar Harbor, US) were maintained in individually ventilated cages (n = 4 per cage), in a controlled environment (21 ± 2 °C; 12 h light/dark cycle; light from 7:00 AM) and fed ad libitum on a standard diet (RM3-P; Special Diet Services (SDS, Horley UK) for 2 weeks ensuring normal development and stabilisation of the microbiota.28 After this time, mice were transferred onto one of two diets, namely control (AIN93-M) or a Neurosyn240 supplemented diet for 12 weeks. The Neurosyn240 diet comprises of the background control diet (AIN93-M) supplemented with 2,157 mg/kg diet of Neurosyn240 (Activ’Inside, Beychac-et-Caillau, France). Neurosyn240 is a proprietary (patent pending) standardized blend of Memophenol™, a unique formula of French Grape (Vitis vinifera L.) and North-American Wild Blueberry (Vaccinium angustifolium A.) extract (patent WO/2017/072219), saffron extract (patent WO/2018/020013), green tea extract, olive leaf extract, trans-resveratrol, zinc, vitamins B5, B9, B12, C, D3 & E (comprising 15% of vitamins and mineral reference value), polyphenols (mainly flavan-3-ol monomers ≥10% and resveratrol ≥1%) and crocin carotenoids (mainly trans-4-galloyl-gallate, trans-3-galloyl-gallocatechin, cis-4-galloyl-gallate, trans-2-galloyl) ≥500 ppm as previously described.29 The dietary fiber content of this mix is at trace amounts and therefore is unlikely to exert an effect upon either the gut microbes or host systems. Diets were prepared by Research Diets Inc. (New Brunswick, USA) to comply with animal nutrition requirements. At the end of the experiments, 5-month-old animals were anaesthetized with a mixture of isoflurane (1.5%) in nitrous oxide (70%) and oxygen (30%) and transcardially perfused with ice-cold saline containing 10 UI/ml heparin (Sigma-Aldrich, UK). Cardiac blood was allowed to clot for 30 mins on ice before sera were isolated via centrifugation at 4,000 x g for 10 min. Brains were rapidly removed, halved, snap frozen and stored at −80 °C until biochemical analysis. Additionally, caeca were removed, weighed and contents were extracted.

All behavioral tests were performed at the experimental endpoint after the 12-week intervention, with mice remaining on the control or Neurosyn240 diet during the testing period. Baseline behavioral testing was not conducted, as prior exposure to the tasks can lead to learning effects, which may confound the interpretation of treatment-related improvements.30 To ensure that observed differences reflect the impact of the intervention rather than task familiarity, animals were naïve to the behavioral tests at the time of assessment. Prior to commencing, a visual placing test was performed on each animal to ensure animals were not visually impaired.31 All behavioral tests were analyzed using the Ethovision software (Tracksys Ltd, Nottingham, UK).

The Open Field (OF) task, a measure of anxiety-related behavior32, was conducted as described previously.33 Animals were individually placed within the (50 cm × 50 cm × 50 cm cubed) arena illuminated with dim lighting (100 lux) and were allowed to freely move for 10 minutes. Mice were tracked using Ethovision software which determined travel distance, velocity and time spent in the center/periphery of the maze respectively.

The novel object recognition (NOR) task, a measure of recognition memory, was conducted as previously described.34^,35 All experiments were carried out under dim lighting conditions (100 lux). Briefly, on day 1 (habituation), mice were placed into an empty arena (50 cm × 50 cm × 50 cm) for 10 minutes. On day 2, animals were conditioned to two identical objects for 10 minutes. Half of the animals were conditioned with two identical golf balls and the other half with two green plastic toy bricks (4x4x6 cm). Following an inter-trial interval of one hour, mice were placed back within the testing arena now containing one familiar object and one novel object, with the position of the novel object (left or right) being randomized between each mouse and group tested. Mice conditioned with the golf balls had the plastic toy brick placed as the novel object and vice versa. Videos were analyzed for a 5-minute period. If the mice did not accumulate a total of 8 seconds of object exploration within this time, the analysis continued for up to 10 minutes or until 8 seconds of exploration was reached. Mice not achieving 8 seconds of exploration were excluded from the analysis.36 Discrimination index (DI) was calculated as follows DI = (TN − TF)/(TN + TF), where TN is the time spent exploring the novel object and TF is the time spent exploring the familiar object.

The Y-maze spontaneous alternation test, a measure of spatial working memory, was conducted as previously described.37 Ethovision software analyzed each animal for 7 minutes recording zone transitioning and locomotor activity. Spontaneous alternation, defined as the tendency of rodents to explore a new arm of the maze rather than returning to one previously visited, reflecting their working memory, was calculated using the formula: (Number of alternations/Total Arm entries - 2) x 100).

The Barnes maze, as previously described,38 was performed with slight modifications to assess spatial retrieval memory. Briefly, the maze consisted of a brightly illuminated (800 lux lighting) circular platform (92 cm diameter), with 20 evenly distributed holes located around the circumference and visual cues (4 simple shapes) placed at the periphery. The experiment was conducted over a 5-day period. On day 1 (habituation) animals were placed into the center of the maze for 2 minutes and were able to explore freely. If the animal did not find the escape box within this time frame the animal was guided to it after which they remained in the box for a further 2 minutes. Following habituation each mouse was tested/trained on their ability to locate the escape box on days 1–4 with three trials per day. On day 5, a probe test was conducted, the maze was rotated 90°, the escape box was removed, and mice were placed in the center of the maze in which they were free to navigate for 1 minute. Percentage time in the correct quadrant was determined using Ethovision software.

Microbial DNA was isolated from approximately 50 mg of cecal contents using a FastDNA SPIN Kit for Soil (MP Biomedicals). Samples were quantified using a nanodrop (ND-1000 spectrophotometer) and quality assessment was performed by agarose gel electrophoresis to detect DNA integrity, purity, fragment size and concentration. The 16 S rRNA amplicon sequencing of the V3–V4 hypervariable region was performed with an Illumina NovaSeq 6000 PE250 using a paired-end 150 bp sequencing strategy. Sequence analysis was performed by Uparse software (Uparse v7.0.1001) 37,39 using all the effective tags. Sequences with ≥97% similarity were assigned to the same operational taxonomic units (OTUs). A representative sequence for each OTU was screened for further annotation. Each sequence was analyzed using Mothur software against the SSUrRNA section of the SILVA database.40 OTUs abundance information was normalized using a standard sequence number corresponding to the sample with the least sequences. Alpha diversity was assessed using Chao1 and Shannon H diversity indices whilst beta diversity was assessed using Bray–Curtis. Statistical significance was determined by Wilcoxon rank-sum test (Mann–Whitney-U) or a permutational multivariate analysis of variance (PERMANOVA).

Liquid chromatography with tandem mass spectrometry (LC-MS/MS) was conducted to gain insight into possible shifts in the production of bioactive serum metabolites and targeted metabolites previously associated with cognitive health, including bile acids, tryptophan, trimethylamine N-oxide (TMAO), p-cresol and its derivatives.41 Serum samples (80 µL) were diluted with methanol at a ratio of 1:10 (v/v) and placed on dry ice for 10 min. Samples were then centrifuged (5 min, 16,000x g at room temp), supernatants filtered using a 0.45 µM PTFE syringe filter and evaporated to dryness using a Savant™ SpeedVac™ High-Capacity Concentrator (Thermofisher, UK). Dried samples were resuspended in either 50 µL of methanol with the addition of 15 µL of lithocholic acid-d4 and cholic acid-d4 at 50 µg/mL for the detection of bile acids, 50 µL water with TMA-d9 N-oxide, TMA ¹³C ¹⁵N hydrochloride at 50 µg/mL for the detection of TMAO/TMA/choline or 50 µL water with 15 µL of L-methionine-3, 3, 4, 4 d4 and p-toluenesulfonic acid at 50 µg/mL for the detection of tryptophan and p-cresol metabolites respectively. All internal standards were supplied from Thermofisher, UK. Samples were analyzed using Waters Acquity UPLC system and Xevo TQ-S Cronos mass spectrometer with MassLynx 4.1 software. For the detection of bile acids, the electrospray ionization (ESI) operated in negative mode and chromatographic separations were performed with a Supelco Ascentis Express C18 column (150 × 4.6 mm, 2.7 µM). A binary gradient of eluent A (10 mM ammonium acetate, 0.1% formic acid, water) and eluent B (10 mM ammonium acetate, 0.1% formic acid, methanol) ran at a constant rate of 0.6 mL/min. For the quantification of TMAO, TMA and choline, ESI operated in positive mode. Chromatographic separation was achieved using an Ethylene Bridged Hybrid (BEH) Amide (150 × 2.1 mm, 1.7 µM) and 0.5 mL/min using eluent A (10 mM ammonium formate, 0.2% formic acid, 50% acetonitrile) and eluent B (10 mM ammonium formate, 0.2% formic acid, 95% acetonitrile). For the detection of tryptophan and p-cresol related metabolites, ESI operated in positive mode using an ACQUITY UPLC BEH C18 1.7 µM (2.1 x 50mm) column. A binary gradient of eluent A (0.1% formic acid, water) and eluent B (0.1% formic acid, methanol) held a constant flow rate of 0.3 mL/min. Chromatogram peak analysis was performed by the accompanying Waters® TargetLynx ™ application manager. Gradients for all three methods are described elsewhere.42

Coronal brain cryosections (20 µm) were processed from snap-frozen hemi-brain samples. Cryosections were mounted onto glass slides before being fixed in 4% formaldehyde in 0.1 M PBS for 15 minutes. To detect amyloid deposits, tissue sections were incubated in 1% thioflavin S aqueous solution (Sigma-Aldrich) for 8 minutes. Sections were further washed twice in 80% ethanol, once in 95% ethanol and thrice in distilled water prior to mounting with Mowiol mounting medium and storage at −20 °C.

To detect the microglial marker protein, Iba-1, sections were incubated for 30 min at room temperature in 10% normal goat serum (NGS) containing 0.05% Triton X-100 to permeabilise cells and block nonspecific secondary antibody binding. Sections were then incubated for 1 hour at room temperature in a rabbit anti-mouse Iba-1 polyclonal antibody (Synaptic Systems) diluted 1/1000 in 1% NGS containing 0.05% Triton X-100. Sections were further washed in 1% NGS and incubated for 1 hour with an AF555-conjugated goat anti-rabbit IgG secondary antibody (diluted 1/500) and washed twice in PBS. 4',6-diamidino-2-phenylindole (DAPI) was used as a counterstain to visualize cell nuclei. Tissue samples were mounted in an aqueous medium, Mowiol, dried and stored at −20 °C.

Images of the hippocampus were collected for the analysis of amyloid deposit pathology and Iba-1 positive microglia using a Direct Fluorescence Leica Ctr 5000 Microscope at x10 magnification. Image analysis was performed manually using Fiji software (NIH, Washington DC, USA). The area of the region of interest (ROI) was outlined manually. Counts of thioflavin S-positive deposits represent the hippocampal region. Iba-1 positive microglia were counted within a randomly selected 200 µm² ROI in the hippocampus.

Total RNA was isolated from hippocampi (n = 3 randomly selected per group and sex) using the Qiazol reagent (Qiagen, UK). Messenger RNA was then purified from total RNA using poly-T oligo-attached magnetic beads. After fragmentation, the first strand cDNA was synthesized using random hexamer primers, followed by the second strand cDNA synthesis using either dUTP for directional library or dTTP for non-directional library.43 The libraries were checked with Qubit and real-time PCR for quantification and bioanalyser for size distribution detection. Quantified libraries were pooled and sequenced on Illumina NovaSeq 6000 platform using a paired-end 150 bp, according to effective library concentration and data amount. Raw data (raw reads) of FASTQ format were first processed through in-house perl scripts. The index of the reference genome was built using Hisat2 v2.0.544 and paired-end clean 1 reads were aligned to the reference genome using Hisat2 v2.0.5. FeatureCounts v1.5.0-p3 was used to count the reads numbers mapped to each gene.45 FPKM of each gene was calculated based on the length of the gene and reads count mapped to this gene.

Data was analyzed using the DESeq2 package in R (version 4.3.2) to investigate the effects of diet and sex on gene expression. Raw count data were filtered to remove low-abundance genes, retaining those with a minimum count of 5 in at least 15% of the samples. Genes with less than 15% variance were excluded. A generalized linear model (GLM) assessed the main effects of diet and sex. Differential expression was tested for each factor, which captures differential responses to diet between sexes. P-values were adjusted using the Benjamini-Hochberg method to control the false discovery rate (FDR). Enrichr was used to perform Gene Ontology (GO) analysis of differentially expressed genes (DEGs).46^,47

RNA was prepared as outlined above. Two μg of total RNA was treated with DNase I (Invitrogen, UK) and used for cDNA synthesis using Invitrogen™ Oligo (dT) primers and M-MMLV reverse transcriptase. Quantitative real-time PCR (qRT-PCR) reactions were performed using SYBR green detection technology on the Applied Biosystems QuantStudio 5 Real-Time PCR system (Life Technologies). Results are presented as relative fold change in Neurosyn240 females and males compared with sex-matched control groups. Values were scaled to the average across all samples for each target gene and normalized to the mean of the reference genes TATA-box binding protein (Tbp) and hypoxanthine phosphoribosyltransferase 1 (Hprt1). The primer sequences are given in Supplementary Table S1.

Data analysis was performed in GraphPad Prism version 9.5.1 (GraphPad Software, CA, USA). All values are presented as means ± standard error of the means (SEM) unless otherwise stated. After identifying outliers using the ROUT method (q = 1%), data were checked for normality using the Shapiro–Wilk test. Analyzes were conducted using two-way ANOVA with Tukey post-hoc analysis when appropriate to correct for multiple comparisons. Weekly body weights for all mice were analyzed using repeated measures two-way ANOVA to assess the effects of diet, sex, and time, as well as their interactions on body weight. Partial Least Squares Discriminant Analysis (PLS-DA) was employed to illustrate the clustering of different metabolites across groups using Metaboanalyst 5.0.48 Dendrogram and heatmaps were created using Spearman's rank correlation for distance measurement and Ward's method for hierarchical clustering. Procrustes analysis, investigating the congruence of the metabolomics and microbiome data, was conducted using M2IA.49 All other correlation analyzes were conducted using Spearman's rank-order correlation analysis. p values of less than 0.05 were considered statistically significant.

Whilst body weight increased over the 12-week intervention for both males and females, animals receiving Neurosyn240 supplementation did not significantly differ from the control group of the same sex (Figure 1A). A main effect of sex was detectable upon body weight (F(1, 28) = 40.4; p < 0.0001), but neither diet (F(1, 28) = 1.16; p = 0.29) nor the interaction between sex and diet (F(1, 28) = 0.05; p = 0.82) reached statistical significance (Figure 1B). There were also no significant differences in food intake throughout the study with both males and females consuming on average 2.9 ± 0.3g per animal, per day, providing an average of 208 ± 21 mg/kg BW/day of Neurosyn240. Using allometric scaling based on body surface area,50 this dose equates to 1.18 ± 0.1 g/day human equivalent dose for a person of 70 kg.

After 12 weeks of dietary intervention, cognitive performance was assessed in 5-month-old 5xFAD mice. There was a near significant effect of diet on time spent in the center of the open field test as a measure of anxiety (F(1, 27) = 4.09; p = 0.06; Figure 2A), but no significant effect of distance traveled in the open field (F(1,27) = 0.04; p = 0.83; Figure 2B), or performance on the Y-maze (F(1,27) = 2.9; p = 0.38; Figure 2C), novel object recognition tests (F(1,27) = 0.097; p = 0.76; Figure 2D) or Barnes maze (F(1,27) = 0.078; p = 0.78; Figure 2E). No significant effects of sex were seen in the open field assessment on time spent in the center (F(1,27) = 0.90; p = 0.35) or distance traveled in the open field (F(1,27) = 2; p = 0.16), as well as Y-maze (F(1,27) = 0.79; p = 0.38), novel object recognition (F(1,27) = 0.0001; p = 0.99) and Barnes maze (F(1,27) = 0.028; p = 0.87). No significant interactions between diet and sex were detected in the behavioral measures.

The impact of Neurosyn240 on microbial composition was investigated by 16S rRNA amplicon sequencing. Alpha diversity, as measured by the Chao1 index, was not significantly affected by sex (F(1,27) = 0.21; p = 0.65) or diet (F(1,27) = 0.53; p = 0.47), nor was there an interaction between sex and diet (F(1,27) = 0.21; p = 0.65). Shannon H index displayed near significant effects of sex (F(1,27) = 0.03; p = 0.06) and diet (F(1,27) = 3.87; p = 0.06) but no significant interaction term (F(1,27) = 0.75; p = 0.39) (Figure 3A). Principal coordinates of analysis (PCoA) highlighted significant differences in beta diversity metrics (PERMANOVA F = 3.15; R² = 0.26; p = 0.001) between males and females in both control and Neurosyn240 groups (Figure 3B). Pairwise comparisons suggested both males (F = 4.39; R² = 0.24; FDR q = 0.018) and females (F = 1.98; R² = 0.13; FDR q = 0.09) had significant differences in beta diversity with the consumption of Neurosyn240 (FDR q < 0.1) (Supplementary Table S2). At the genus level, linear discrimination analysis (LDA) effect size (LEfSe) revealed significant differences in 16 genera between groups (FDR q < 0.05; Supplementary Figure S1). Of these bacteria, a main effect of diet was detected in Lactococcus (F(1,27) = 6.67; p = 0.016), Limosilactobacillus (F(1,27) = 12.12; p = 0.002), Marvinbryantia (F(1,27) = 17.22; p < 0.001), Parvibacter (F(1,27) = 25.52; p < 0.001), Romboutsia (F(1,27) = 12.69; p = 0.001), Sporosarcina (F(1,27) = 10.88; p = 0.003), Turicibacter (F(1,27) = 4.30; p = 0.048), Eubacterium_fissicatena_group (F(1,27) = 8.77; p = 0.006), Bifidobacterium (F(1,27) = 4.26; p = 0.049) and Dubosiella (F(1,27) = 13.14; p = 0.001) (Figure 3C; Supplementary Table S3 for all genera abundances). Eight genera were significantly influenced by the main effect of sex, including Lactococcus (F(1,27) = 4.28; p = 0.048), Limosilactobacillus (F(1,27) = 12.12; p = 0.0017), Marvinbryantia (F(1,27) = 7.407; p = 0.011), Sporosarcina (F(1,27) = 10.170; p = 0.004), Turibacter (F(1,27) = 15.22; p = 0.001), Enterococcus (F(1,27) = 6.766; p = 0.015), Enterorhabdus (F(1,27) = 5.444; p = 0.027) and Acetatifactor (F(1,27) = 6.77; p = 0.015). Significant interaction effects were observed for Lactococcus (F(1,27) = 5.43; p = 0.028), Limosilactobacillus (F(1,27) = 12.34; p = 0.0016), Enterococcus (F(1,27) = 7.008; p = 0.013), Marvinbryantia (F(1,27) = 8.424; p = 0.007), Romboutsia (F(1,27) = 4.163; p = 0.051), Sporosarcina (F(1,27) = 13.940; p = 0.001) and Turicibacter (F(1,27) = 4.296; p = 0.048).

Targeted metabolomic profiling was conducted to gain insight into possible shifts in the production of bioactive metabolites associated with cognitive health. PLS-DA analysis showed a shift in metabolic response to the Neurosyn240 diet in both males and females (Figure 4A). This was further supported by heatmap analysis, which demonstrated alterations in the relative abundance of 33 metabolites in response to the Neurosyn240 diet compared to the control in males and females (Figure 4B). Seven metabolites were significantly impacted by the main effect of diet, including serotonin (F(1, 27) = 13.14; p = 0.001; Figure 4C), kynurenine (F(1,27) = 6.46; p = 0.018; Figure 4D), taurocholic acid (TCA; F(1, 24) = 11.44; p = 0.002; Figure 4E), hyodeoxycholic acid (HDCA; F(1, 27) = 4.51; p = 0.044; Figure 4F), taurodeoxycholic acid (TDCA; F(1, 27) = 5.65; p = 0.025; Figure 4G), chenodeoxycholic acid (CDCA; F(1, 27) = 8.41;p = 0.008; Figure 4H) and lithocholic acid (LCA; F(1, 27) = 11.11; p = 0.003; Figure 4I). See Supplementary Table S4 for all metabolite concentrations. To assess the relationship between dietary modulation of the gut microbiome and the circulatory metabolome, Spearman rank correlation analysis was performed on metabolites and microbiota genera modulated by the main effect diet (p < 0.05). This analysis revealed a significant correlation between the peripheral metabolome and the gut microbiome (Figure 4J). Procrustes analysis was conducted to evaluate the congruence of the two datasets. Significant similarities were observed between males (R = 0.69; p < 0.001; Supplementary Figure S2A) and females (R = 0.64; p < 0.001; Supplementary Figure S2B) on control and Neurosyn240 diet, suggesting similarity between microbiome and metabolome profiles.

A main effect of sex was observed on numerous tryptophan-related metabolites, including tryptophan (F(1,27) = 35.07; p < 0.001), 5-hydroxindole-acetic acid (5HIAA; F(1, 27) = 11.93; p = 0.002), anthranilic acid (F(1,27) = 10.43; p = 0.004), kynurenic acid (F(1,27) = 5.20; p = 0.031), indole acetic acid (IAA; F(1,27) = 89.41; p < 0.001), indole-3-lactic acid (ILA; F(1,27) = 10.12; p = 0.004), indole-3- carboxaldehyde (I3A; F(1,27) = 8.69; p = 0.007), indole (F(1,27) = 12.72; p = 0.002), indoxyl sulfate (F(1,27) = 10.28; p 0.004) (Supplementary Figure S3). Trimethylamine-N-oxide (TMAO; F(1,27) = 23.92; p < 0.001), THDCA; (F(1,24) = 7.14; p = 0.013), cholic acid (CA; F(1,27) = 7.01; p = 0.014), HDCA (F(1,27) = 30.39; p < 0.001), DCA; F(1,27) = 32.58; p < 0.001), TCA (F(1,27) = 21.24; p < 0.001) and TDCA (F1,27) = 4.74; p = 0.039) were also significantly influenced by sex (Supplementary Table S4). A significant interaction between sex and diet was observed in 5HIAA (F(1,27) = 5.52; p = 0.027), p-cresol sulfate (F(1,27) = 9.18; p = 0.006), p-cresol glucuronide (F(1,27) = 26.31; p < 0.001), alpha-muricholic acid (α-MCA; F(1,27) = 5.14; p = 0.032), CA (F(1,27) = 12.44; p = 0.002), ursodeoxycholic acid (UDCA; F(1,27) = 5.42; p = 0.028), HDCA (F(1,27) = 12.47; p = 0.002), DCA (F(1,27) = 6.47; p = 0.018), TCA (F(1,27) = 17.51; p = 0.003) and TDCA (F(1,27) = 6.52; p = 0.017) (Supplementary Table S4).

In the hippocampus, there was a significant effect of diet on the percentage area of amyloid deposits (F(1,27) = 4.33; p = 0.048; Figure 5A) and a near significant effect of diet on the number of amyloid deposits (F(1,27) = 3.68; p = 0.06; Figure 5B-C). However, there was no effect of sex on the percentage of amyloid deposits (F(1,27) = 2.20; p = 0.15) or number of amyloid deposits (F(1,27) = 1.06; p = 0.31). Similarly, no interaction was observed between sex and diet on the percentage area of amyloid deposits (F(1,27) = 0.83; p = 0.37) or number of amyloid deposits (F(1,27) = 0.35; p = 0.56). qPCR analysis showed no significant effects of diet on hippocampal expression of APP-processing genes, including presenilin 1 (Psen1; F(1,27) = 0.12; p = 0.73), beta-site APP-cleaving enzyme 1 (Bace1; F(1,27) = 0.37; p = 0.55), amyloid beta precursor protein (App; F(1,27) = 0.04; p = 0.85), and ADAM metallopeptidase domain 10 (Adam10; F(1,27) = 0.71; p = 0.41) (Figure 5D–G). There were no significant effects of sex or sex × diet interactions (p > 0.05).

Given the absence of a significant main effect of sex or interaction between sex and diet, correlation analysis was conducted on all mice as a single group to examine the relationship between metabolites significantly modulated by diet (p < 0.05) and amyloid deposits (Supplementary Figure S4). A significant positive association was found between serum LCA concentrations and the number of amyloid deposits in mice (r = 0.39; p = 0.04), suggesting that these changes may be linked to metabolite levels (Figure 5H).

In the hippocampus, a significant main effect of diet was observed on the number of Iba-1 positive microglia (F(1,27) = 8.55; p = 0.007; Figure 6A-B), but no effect of sex (F(1,27) = 1.31; p = 0.26) or interaction between sex and diet (F(1,27) = 2.20; p = 0.15). Consistent with this, qPCR analysis revealed a significant effect of diet on the expression of markers of disease-associated microglia (DAM), triggering receptor expressed on myeloid cells 2 (Trem2; F(1,27) = 5.49; p = 0.03) and a near significant effect on TYRO protein tyrosine kinase binding protein (Tyrobp; F(1,27) = 3.78; p = 0.06) (Figure 6C-D). Furthermore, there was a significant effect of diet on the proinflammatory cytokine tumor necrosis factor (Tnf) expression (F(1,27) = 5.96; p = 0.02), a near-significant effect on interleukin 1 beta (Il1b; F(1,27) = 3.79; p = 0.062), and no significant effect on interleukin 6 (Il6; F(1,27) = 2.77; p = 0.11) (Figure 6E–G). There were no significant effects of sex or sex × diet interactions (p > 0.05). The number of hippocampal Iba-1 positive microglia negatively correlated with circulatory serotonin concentrations (r = −0.38; p = 0.045), suggesting a possible gut-brain interaction (Figure 6H).

In addition to microglial markers, hippocampal expression of genes linked to neuronal health and AD pathology was assessed. VGF Nerve Growth Factor Inducible (Vgf) and the dual-specificity phosphatases Dual Specificity Phosphatase 4 (Dusp4) and Dual Specificity Phosphatase 6 (Dusp6), which have been implicated in amyloid clearance and memory function, were measured by qPCR.51-53 There was no effect of diet, sex or interaction between sex and diet for DUSP4 (F(1,27) = 0.76; p = 0.39), DUSP6 (F(1,27) = 0.55; p = 0.47) or Vgf (F(1,27) = 0.55; p = 0.47) (Supplementary Figure S5).

To examine the potential neuroprotective mechanisms of Neurosyn240, RNA sequencing analysis was conducted on the hippocampal brain region, a region significantly affected by AD.54 Using DESeq2 analysis, we identified 47 DEGs that were upregulated and 77 genes that were downregulated by diet (Figure 7A). Additionally, 7 genes were upregulated, and 7 genes were downregulated by sex. Hierarchical clustering analysis displayed the expression of significant DEGs (Figure 7B). GO analysis highlighted the functional properties of DEGs. Neurosyn240 upregulated processes related to neuronal health, including positive regulation of biosynthetic processes (GO:0009891) and dopaminergic neuron differentiation (GO:0071542), critical for maintaining neuronal function (Figure 7C).55^,56 Notably, pathways related to calcium-mediated signaling (GO:0019722, GO:0035584) and protein localization to cell junctions (GO:1902414) were significantly upregulated, suggesting synaptic signaling and cellular communication may be improved (P_FDR < 0.1). Several key inflammatory and immune-related pathways were downregulated by Neurosyn240 consumption (p < 0.05). Additionally, NF-kappaB signaling (GO:1901224), a central regulator of inflammation, along with pathways involved in the positive regulation of cytokine production (GO:1900017), were downregulated. However, it should be noted that these pathways did not reach significance at an adjusted p-value < 0.1.

GO analysis revealed sexually dimorphic effects on histone modifications and signaling pathways (Figure 7D). Females showed upregulation of H3-K4 and histone lysine methylation, downregulation of histone lysine demethylation, and reduced androgen and steroid hormone receptor signaling compared to males, aligning with previous studies.57^,58

To investigate potential pathways underlying hippocampal reductions in amyloid deposits, gene expression results were compared to a curated set of 155 genes associated with amyloid beta formation, binding, and clearance, identifying those consistently present across two functional genome databases.59-61 DESeq2 GLM analysis showed that the Neurosyn240 significantly (P_FDR < 0.1) upregulated expression of genes associated with amyloid beta binding (Clusterin (Clu); P_FDR = 0.051), clearance (Low-Density Lipoprotein Receptor-Related Protein 2 (Lrp2); P_FDR = 0.045) and cellular response (Vascular Cell Adhesion Molecule 1 (Vcam1); P_FDR = 0.027), while downregulating the amyloid binding gene Ager P_FDR = 0.07 (Table 1; Supplementary Table S5 for full list of gene expression). No genes associated with amyloid beta formation, binding or clearance were significantly modulated by sex (Supplementary Table S6). Correlating genes and circulatory metabolites significantly modulated by diet highlighted a negative correlation between Lrp2 and LCA concentrations (r = −0.60; p = 0.041; Supplementary Figure S6), suggesting a potential link between the metabolome and the brain.