Supplement of microbiota-accessible carbohydrates prevents neuroinflammation and cognitive decline by improving the gut microbiota-brain axis in diet-induced obese mice

Sixty C57Bl/6 J male mice (11 weeks old) were obtained from the Experimental Animal Center of Xuzhou Medical University (Xuzhou, China, SCXK (Su)20150009), and housed in environmentally controlled conditions (temperature 22 °C, 12 h light/dark cycle). After a 1week acclimatization period, mice were used for experiments in accordance with the Chinese Council on Animal Care Guidelines and approved by the Institutional Animal Care Committee of Xuzhou Medical University.

The mice were randomly divided into three groups (n = 15): (1) the control group (Con) was fed with a grain-based rodent lab chow (LabDiet 5010, 50 g/kg from fat, (5% fat by weight); plant polysaccharide-rich ranging a diverse source of plants including corn, soybean, wheat, oats, alfalfa, and beet, (15% neutral detergent fiber by weight)); (2) the HF-FD group was fed with a diet with high-fat (315 g/kg from fat: soybean oil 55 g and lard 260 g, (31.5% fat by weight)) and fiber-deficient (50 g/kg cellulose, low accessibility by gut microbiota, (5% fiber by weight)); (3) the group was fed with a diet rich in MACs based on the HF-FD (HF-MAC): mixed with 316 g/kg from fat (soybean oil 56 g and lard 260 g, (31.6% fat by weight)), and LabDiet 5010 powder 634 g/kg (obtained from plant polysaccharide-rich ranging a diverse source of plants including corn, soybean, wheat, oats, alfalfa, and beet, 10% neutral detergent fiber by weight) as previously described [24] (diet details are outlined in Table S1). In addition, another two groups (n = 12) paralleled with HF-FD group and HF-MAC group were given a cocktail of antibiotics (HF-FD + AB, HF-MAC + AB) in the drinking water to investigate the role of gut microbiota in MAC intervention. Two days prior to HF-FD diet initiation, the HF-FD + AB and HF-MAC + AB mice were given drinking water containing ampicillin (1 g/L), vancomycin (0.25 g/L), neomycin (1 g/L), and metronidazole (1 g/L), which was prepared fresh every 3 days [25]. The HF-FD diet group showed increased body weight compared to the control group from week 4 onwards, in addition to exhibiting a significantly higher final body weight following completion of the 15-week diet (32.6%, p < 0.01). The supplementation of MACs in the HF-FD diet suppressed body weight gain and exhibited a lower final body weight than HF-FD mice (− 8.13%, p < 0.05); however, they were still significantly higher than that of the control group (18.9%, p < 0.05, Fig. S1A). Furthermore, fat accumulation (Fig. S1B) and liver weight (Fig. S1C) were significantly increased in the HF-FD mice compared to control mice. In addition, they displayed adipocyte hypertrophy (Fig. S1D), crown-like structures (CLS) in which macrophages surround dead adipocytes (Fig. S1E), signs of hepatic steatosis (ballooning and steatosis) (Fig. S1F and G), altered glucose metabolism, including hyperinsulinemia, increased homeostatic model assessment-insulin resistance (HOMR-IR), and glucose tolerance (Fig. S1H-K). MAC supplementation to some degree attenuated these metabolic disturbances. All the groups were given intervention for 15 weeks followed by two cognitive behavior tests (described below).three days later, the mice were sacrificed by CO2 asphyxiation and blood, cecum content, colon, and brain tissue were collected for further examination.

The nesting behavior and temporal order memory tests were performed as previously described [26, 27], to evaluate spontaneous rodent behavior and recognition memory, respectively. In brief, during the nesting behavior test, the deacon nest score and untorn nestlet weight were used to evaluate the activities of daily living typically altered in patients with cognitive impairment.

The temporal order memory test comprised two sample trials and one test trial with an inter-trial interval of 60 min between each trial. In each sample trial, the mice were allowed to explore two copies of the same object for 4 min; however, the objects were different between the two sample trials (sample trial 1: object A and A’; sample trial 2: object B and B’). During the test trial, one object from sample trial 1 (A; old familiar) and another object from sample trial 2 (B; recent familiar) were presented, and the animals were allowed to explore the open field for 3 min. A discrimination ratio was calculated by using the formula [(old familiar time − recent familiar time)/total exploration time]. Intact object recognition memory for temporal order was considered if the mice spent more time exploring the old familiar object compared with the recent familiar object.

Genomic DNA amplification and sequencing were conducted as our previous study [5]. Briefly, microbial DNA was extracted from the cecal contents of mice using the E.Z.N.A. stool DNA Kit (Omega Bio-tek, Norcross, GA, U.S.) according to manufacturer’s protocols. The 16S rDNA V3-V4 region of the eukaryotic ribosomal RNA gene was amplified by PCR (95 °C for 2 min, followed by 27 cycles at 98 °C for 10 s, 62 °C for 30 s, and 68 °C for 30 s and a final extension at 68 °C for 10 min) using primers 341F:CCTACGGGNGGCWGCAG; 806R:GGACTACHVGGGTATCTAAT, where the barcode is an eight-base sequence unique to each sample. PCR reactions were performed in triplicate, with a 50 μL mixture containing 5 μL of 10 × KOD buffer, 5 μL of 2.5 mM dNTPs, 1.5 μL of each primer (5 μM), 1 μL of KOD polymerase, and 100 ng of template DNA. Amplicons were extracted from 2% agarose gels, purified using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA), and quantified using QuantiFluor-ST (Promega, USA). Purified amplicons were pooled in equimolar and paired-end sequenced (2 × 250) on an Illumina platform according to the standard protocols.

For quantification of total fecal bacterial load, total DNA was isolated from known amounts of feces using the QIAamp DNA Stool Mini Kit (Qiagen). DNA was then subjected to quantitative PCR using the QuantiFast SYBR Green PCR Kit (Biorad) with universal 16S rRNA primers (5′-AGAGTTTGATCCTGGCTCAG-3′ and 5′-CTGCTGCCTCCCGTAGGAGT-3′) to measure total bacteria number. Results are expressed as bacteria number per mg of stool, using a standard curve.

Fecal pellets were collected prior to the behavioral tests, snap frozen in liquid nitrogen, and stored at − 80 °C. Pellets were resuspended at 10 mg/ml in sterile phosphate buffered saline (PBS) and the concentration of albumin determined by ELISA in accordance with manufacturer’s instructions (Elabscience, Cat.E-EL-M0656c, China).

Serum TNF-α, interleukin-6 (IL-6), and interleukin-1β (IL-1β) levels were measured using ELISA kits and performed according to the manufacturer’s instructions (Thermo Fisher, USA). GC–MS analysis of short-chain fatty acid composition in the serum was performed as previously described [28]. Briefly, separation was performed on an Agilent HP-INNOWAX capillary column (30 m × 0.25 mm × 0.25 μm). The column temperature was held at 90 °C for 1 min, increased to 120 °C at 10 °C/min, held for 8 min, then increased to 150 °C at 5 °C/min, before being increased to 250 °C at 25 °C/min, and held for 2 min. The injection volume was 1.0 μL with a split ratio 10:1. The carrier gas was high-purity helium with a flow rate of 1.0 mL/min. The mass spectrometer was operated in electron impact mode (70 eV) at 0.2 s/scan and recorded over the mass range of m/z 50–500, a solvent delay time of 2 min. The inlet, interface, and ionization source temperatures were 250 °C, 230 °C, and 250 °C, respectively.

The concentration of circulating serum LPS was measured by ELISA (Limulus assay kit, Cat.18110115, China). All samples for LPS measurements were performed in duplicate.

The IPGTT was conducted as previously described by our laboratory [29]. Briefly, mice were fasted overnight followed by an intraperitoneal injection of glucose (2 g/kg). Blood samples were obtained from the tail vein at 0, 30, 60, 90, and 120 min following the injection of glucose. Blood glucose levels were measured with glucose meter. Total area under the curve (AUC) was calculated using the trapezoidal method.

Fasting blood glucose was measured using glucometer strips (Roche, Germany). Fasting insulin was measured using a commercial ELISA kit (Crystal Chem, USA). Homeostatic model assessment-insulin resistance (HOMA-IR) was calculated using the following equation: fasting insulin (mU·L − 1) × fasting blood glucose (mmol·L − 1)/22.5 [29].

Post Carnoy’s fixation, methanol-stored colon samples were embedded in paraffin, cut into thin sections (5 μm), and mounted on glass slides. Alcian blue staining was performed as previously described [30], and the thickness of the colonic sections was measured (10 measurements per section/2 sections per animal/5 animals per group) using ImageJ after cross-validation using anti-MUC2 staining.

The staining of bacteria localization at the surface of the intestinal mucosa was conducted by FISH as previously described [31]. Briefly, transverse colonic tissues full of fecal material were placed in methanol-Carnoy’s fixative solution (60% methanol, 30% chloroform, 10% glacial acetic acid) for a minimum of 3 h at room temperature. Tissues were then washed in methanol (2 × 30 min), ethanol (2 × 20 min), and xylene (2 × 20 min). Paraffin-embedded tissue was cut into sections (5 μm). The tissue sections were dewaxed by preheating at 60 °C for 10 min, followed by incubation in xylene (2 × 20 min) and 100% ethanol for 10 min. Deparaffinized sections were incubated at 37 °C overnight with EUB338 probe (5′-GCTGCCTCCCGTAGGAGT-3′) diluted to 10 μg/mL in hybridization buffer (20 mM Tris–HCl, pH 7.4, 0.9 M NaCl, 0.1% SDS, 20% formamide). Subsequently, sections were incubated with wash buffer (20 mM Tris–HCl, pH 7.4, 0.9 M NaCl) for 10 min and PBS for 3 × 10 min, before mounted in DAPI containing mounting medium.

Mucin 2 (MUC2) localization in the colon was detected by staining the colonic tissue sections (5 μm) with anti-MUC2 antibody (Abclonal, A14659) diluted 1:500 in TBS, followed by incubation with goat-anti-rabbit Alexa-488 conjugated antibody (1:1000) (Invitrogen, A32731) in TBS. Frozen brain tissue (hippocampus) was cut in 20 μm sections using a cryostat from bregma − 3.3 mm to − 4.16 mm according to a standard mouse brain atlas [32]. The brain slices were incubated with 10% goat normal serum for 15 min at room temperature, followed by incubation with the primary antibodies, anti-Iba1 (Wako, 019–19,741), anti-CD68 (BIO-RAD, MCA1957T), and anti-GFAP (abcam, ab7206) at 4 °C overnight. Following primary antibody incubation, sections were washed with PBS and incubated with either Alexa Fluor®-594 (abcam, ab150160) or Alexa Fluor®-488 (abcam, ab150077) at 37 °C for 1 h. Finally, the sections were counterstained with DAPI (Sigma, D9542) and then imaged with microscope (OLYMPUS IX51). Quantification of positively stained cells in the CA1, CA3, and DG regions were counted using ImageJ. Images obtained from the CA1, CA3, and DG regions were thresholded and processed by a cleaning algorithm including size-based particle exclusion and manual pruning of overlapping cell profiles. For each subsequent profile, the morphological parameters of area and perimeter were calculated. Circularity was calculated by the following formula: 4π × (area/perimeter²). Ramification index of Iba1⁺ and GFAP⁺ cells were quantified by grid-cross analysis using the ImageJ 1.46r (http://imagej.nih.gov/ij/download.html↗) [33, 34]. Briefly, a grid was overlaid by images of the CA1, CA3, and DG regions. The number of grid-crossing points per individual cell was counted, and the mean number of grid-crossed points per cell was calculated.

Epididymal adipose tissue was fixed in 10% buffered formaldehyde and embedded in paraffin. Tissue sections (5 μm) were stained with hematoxylin and eosin and imaged at × 100 magnification under OLYMPUS microscope (BX51, Japan). Using the software ImageJ 1.46r, 2 fields per section and 6 sections per fat mass were analyzed to quantify the number of crown-like structures (CLS) consisting of dead adipocytes surrounded by macrophages.

To determine the degree of liver damage, fresh frozen liver sections (10 μm) were stained with hematoxylin and eosin. Three fields per section in 3 sections of each mouse were viewed under an OLYMPUS microscope (BX51, Japan) and images were captured. The histological parameters of steatosis and ballooning were scored as previously described [29]. The steatosis grades were defined as follows: 0 (< 5%), 1 (5–33%), 2 (> 33–66%), and 3 (> 66%). The ballooning classifications were grouped as 0, no ballooning cells; 1, few ballooning cells; 2, many cells/prominent ballooning.

Total RNA was extracted from homogenized tissues in Trizol (Thermo Fisher Scientific, Waltham, MA, USA) under sterile conditions. Purified RNA (1 μg) was used for RT-PCR to generate cDNA with a High-Capacity cDNA Reverse Transcription Kit (Takara, Dalian, China) The resulting cDNA was used for quantitative PCR in a real-time PCR detection system (Bio-Rad, Hercules, CA, USA). The relative mRNA expression level was determined with the 2-ΔΔCt method with GAPDH as the internal reference control. Primer sequences were as follows: mTNFα--forward (F): CTTGTTGCCTCCTCTTTTGCTTA, mTNFα--reverse (R): CTTTATTTCTCTCAATGACCCGTAG, mIL-1β--forward (F): TGGGAAACAACAGTGGTCAGG, mIL-1β--reverse (R): CTGCTCATTCACGAAAAGGGA, mIL-6--forward (F): TCACAGAAGGAGTGGCTAAGGACC, mIL-6--reverse (R): ACGCACTAGGTTTGCCGAGTAGAT, mReg3γ--forward (F): 5′TTCCTGTCCTCCATGATCAAA-3′, mReg3γ--reverse (R):5′CATCCACCTCTGTTGGGTTC-3. mGAPDH--forward (F): AGAAGGTGGTGAAGCAGGCATC, mGAPDH--reverse (R): CGAAGGTGGAAGAGTGGGAGTTG. mCD68--forward (F): TCACCTTGACCTGCTCTCTCTAA, mCD68--reverse (R): GCTGGTAGGTTGATTGTCGTCTG.

Mouse colon and hippocampal tissue were homogenized in ice-cold RIPA lysis buffer, supplemented with complete EDTA-free protease inhibitor cocktail and PhosSTOP phosphatase inhibitor. The homogenate was sonicated 6 times for 4 s with 6-second interval on ice and then centrifuged at 12,000 g for 20 min at 4 °C. The supernatant was collected, and the protein concentration was quantitated using a BCA assay. Equal amounts of protein were separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes. Subsequently, membranes were blocked with 5% non-fat milk at room temperature for 1 h and then incubated with the primary antibodies at 4 °C overnight. Primary antibodies included anti-Occludin (Abcam, ab167161), anti-ZO1 (Abcam, ab96587), anti-p-IRS-1 (Ser307) (CST, 2381 T), anti-IRS-1 (CST, 2382S), anti-Iba1 (Wako, 019–19,741), anti-GFAP (Abcam, ab7206), anti-p-GSK-3β(Ser9) (CST, 9322S), anti-GSK-3-β (CST, 12456 T), anti-p-AKT (Ser473) (CST, 4060 T), anti-AKT (CST, 4691 T), anti-Tau5 (Abcam, ab80579), anti-p-Tau (S202 + T205) (Abcam, ab80579), anti-Synaptophysin (Abcam, ab32127), anti-PSD95 (CST, 3450), anti-Synapsin I (Abcam, ab18814), anti-PTP1B (Abcam, ab189179), GAPDH (ABclonal, AC033), and β-Actin (ABclonal, AC026). Following primary antibody incubation, membranes were washed in TBST and incubated with either HRP-linked anti-rabbit IgG secondary antibody (CST, 7074) or HRP-linked anti-mouse IgG secondary antibody (CST, 7076S) at room temperature for 1 h. Protein bands were detected with Clarity™ ECL Western Blot substrate (Bio-Rad, 1,705,060) and visualized using ChemiDoc Touch imaging system (Bio-Rad).

Mice were sacrificed by CO2 asphyxiation and then transcardially perfused with 4% paraformaldehyde. The hippocampal CA1 region was collected and rapidly fixed in a solution composed of 4% paraformaldehyde and 2.5% glutaraldehyde, following a wash with 0.1 M phosphate buffer solution (PBS, pH 7.4) and postfixed with 1% osmic acid for 2 h. Subsequently, the tissue was washed with double distilled water and dehydrated with an ethanol and acetone gradient. Samples were embedded with different concentrations of epoxy resin and polymerized at 37 °C for 24 h, followed by 45 °C for 24 h, and 60 °C for 24 h. The samples were cut into ultrathin sections (70 nm) and stained with uranyl acetate and lead citrate. Two grids per specimen and 10 photographs per grid were randomly taken of the synaptic terminals and viewed on a transmission electron microscope (TEM) (FEI Tecnai G2 Spirit TWIN, America) to estimate synaptic morphometry. Gray type I synapses (asymmetric synapses considered to mediate excitatory transmission) were identified in the micrographs by the presence of synaptic vesicles (SVs) and dense material in postsynaptic axon terminal. The postsynaptic density (PSD) thickness was evaluated as the length of a perpendicular line traced from the postsynaptic membrane to the most convex part of the synaptic complex. The widths of the synaptic clefts (SCs) were estimated by measuring the widest and narrowest portions of the synapse and then averaging these values.

Data was analyzed using the statistical package SPSS (version 20, IBM Corporation, Chicago, IL, USA). Data was firstly tested for normality before differences among the control and HF-FD and HF-MAC groups were assessed using one-way analysis of variance (ANOVA), followed by a post hoc Tukey’s Honest Significant Difference (HSD) test or Kruskal-Wallis test for multiple comparisons among the groups. p < 0.05 was considered to be statistically significant. For 16S rRNA gene sequence analysis, all reads were deposited and grouped into operational taxonomic units (OTU) at a sequence identity of 97% [35], and the taxonomic affiliation of the OTUs was determined with quantitative insights into microbial ecology (QIIME, version 1.8.0) against the Greengenes database (version 13.8) [36]. Based on Kyoto Encyclopedia of Genes and Genomes (KEGG) functional pathway, the predicted functional composition of the intestinal microbiome was inferred for each sample using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) [37]. Statistical analyses were conducted with STAMP [38], and functional differences in orthologs among groups were assessed by a one-way ANOVA followed by post hoc Tukey-Kramer or Kruskal-Wallis test for multiple comparisons.

To investigate the effects of high fat and fiber deficiency on intestinal microbiota and whether microbiota-accessible carbohydrate could prevent microbiota dysbiosis, qRT-PCR was used to evaluate total fecal bacterial loads. Relative to control mice, the HF-FD diet fed mice resulted in an approximate 10-fold reduction in total fecal bacterial loads (p < 0.05). Furthermore, this reduction in fecal bacterial load was restored by MAC supplementation (Fig. 1a), suggesting that MACs support a normal level of bacterial growth. Moreover, the richness, diversity, and composition of gut microbiota were examined by 16S rRNA sequencing. MACs significantly increased richness as assessed by the Chao1 and ACE index (Fig. 1c, d) and α-diversity in the Shannon index, (Fig. 1e), which were significantly decreased by HF-FD diet. Principal component analysis (PcoA) of the bray distance for β-diversity showed a clear separation between the MAC supplementation group and HF-FD group (Fig. 1b). At the phyla level, the abundance of Firmicutes and Proteobacteria was significantly increased in HF-FD group compared to control group (Fig. 1f), while MAC supplementation decreased Proteobacteria, but not Firmicutes (Fig. 1f). The relative abundance of Bacteroidetes was dramatically decreased in HF-FD group, however, was increased by MAC supplementation (Fig. 1f).

When comparing the microbiota of control, HF-FD and HF-MAC mice using an LDA effect size (LEfSe) calculation, we found 24 differentially abundant taxonomic classes (p < 0.05) with an LDA score higher than 2.0 (Fig. 1h). The results showed that phylum Proteobacteria, class Deltaproteobacteria, order Desulfovibrionales, and family Desulfovibrionaceae were all significantly increased in HF-FD mice, compared to control (Fig. 1h). While mice in the MAC supplementation group showed significantly increased genus Streptococcus, genus Faecalibacterium, genus Prevotella, phylum Cyanobacteria, class 4C0d_2, order YS2, class Coriobacteriia, order Coriobacteriales, family Coriobacteriaceae, phylum Actinobacteria, class Actinobacteria, order Bifidobacteriales, family Bifidobacteriaceae, genus Bifidobacterium, and species pseudolongum (p < 0.05, Fig. 1h, i).

Moreover, KEGG results showed that among the 3 dietary groups, 9 enrichment scores were different in functionally enriched KEGG pathways at level two within 5 level one in the microbiota community (Table 1). Totally, 6 functional orthologs were significantly altered in HF-FD group in the level two KEGG pathways, when compared to the control group. The MAC supplementation was associated with marked 7 microbial functional shifts, including signal transduction, signaling molecules and interaction, cell motility, neurodegenerative diseases, metabolic diseases, environmental adaption and digestive system.

Following our observation that MACs prevented gut microbiota dysbiosis, we examined the effects of MACs intake on the integrity of the colonic barrier. The thickness of colonic mucus was significantly increased in MAC supplementation group compared with HF-FD group, assessed by alcian blue staining (Fig. 2a, b) and mucin-2 glycoprotein (MUC2) immunofluorescence staining (Fig. 2c). Furthermore, the microbiota-epithelial localization in the colon was examined by FISH (Fig. 2d). MACs increased the distance between microbiota and epithelial cells, which was shorter in HF-FD group, indicating microbiota encroachment was significantly reversed by MAC supplementation. Furthermore, we observed that antimicrobial peptide Reg3γ mRNA expression was significantly increased in the colon tissue (Fig. 2e), suggesting MACs may increase the ability for the mucosa to protect against bacterial infection. In addition, the levels of tight junction proteins, occludin, and zonula occludens-1 (ZO-1), were significantly increased in the colon of the MAC supplementation group (both p < 0.05, Fig. 2f, g) with decreased fecal albumin and LPS in serum (Fig. 2h, i), indicating enhancement of gut barrier integrity attenuated gut permeability.

Following the ability of MACs to prevent gut barrier interruption and endotoxinemia, we further investigated the effects of MAC supplementation on intestinal and systemic inflammation and serum SCFA level. Firstly, the colon length was significantly increased by MAC supplementation (p < 0.05, Fig. 3a). As colon length is negatively associated with inflammation [39], we consequently confirmed that the mRNA levels of pro-inflammatory cytokines, TNFα, IL-6, and IL-1β, were significantly decreased in the colon of HF-MAC group compared with HF-FD group (p < 0.05, Fig. 3b–d). Furthermore, the serum levels of these pro-inflammatory cytokines were reduced by MAC supplementation compared with the HF-FD diet group (p < 0.05, Fig. 3 e, f). To demonstrate that MACs are utilized by the gut microbiome to generate SCFAs, the serum levels of SCFAs were measured. HF-MAC supplementation group showed increased levels of acetic acid, propionic acid, and butyric acid in the serum compared to the HF-FD diet group (p < 0.05, Fig. 3h–j). Collectively, these results suggest that MAC supplementation prevented colonic and systemic inflammation induced by obesogenic HF-FD diet.

It is reported that LPS induces blood-brain barrier (BBB) disruption [40], which has been associated with several neurodegenerative diseases, including Alzheimer’s disease [41]. Furthermore, loss of tight junctional proteins has been linked to an increased BBB permeability [42]. We found that occludin protein levels in the hippocampus were significantly decreased in HF-FD group, while MAC supplementation restored this deficit (p < 0.05, Fig. 4a). Activation of neuroglia is implicated in the pathogenesis of memory decline in neuroinflammatory and neurodegenerative diseases [1]. Therefore, we examined the protein level of the microglia and astrocyte markers, Iba1 and GFAP, respectively, in the hippocampus, following response to the three types of diets. Western blot analysis showed the HF-FD group exhibited increased Iba1 and GFAP protein expression (p < 0.05, Fig. 4b and g), which was attenuated by MAC supplementation. The number and morphology of microglia and astrocytes presented differently among the three groups examined by immunofluorescent staining with Iba1, CD68, and GFAP antibodies, respectively (Fig. 4c–j). In the HF-FD group, the total number of Iba1⁺ microglia increased. Furthermore, morphological analysis of Iba1⁺ cells in the HF-FD group resembled activated microglia, evidenced by increased circularity index and decreased ramification index within the CA1, CA3, and DG of hippocampus (Fig. 4d and e). Moreover, the HF-MAC group, exhibited reduced number of Iba1⁺ cells, decreased circularity index and increased ramification index in the CA1, CA3, and DG compared to the HF-FD group. In addition, the HF-FD group exhibited elevated mRNA levels of the activated microglial marker, CD68, and increased number of CD68⁺ cells (Fig. 4f, g) compared to the control group and HF-MAC group. Immunohistochemical analysis of GFAP⁺ cells revealed an increased number of GFAP⁺ in the CA1, CA3, and DG of hippocampus in HF-FD group compared to the control group. In line with this, GFAP⁺ cells in the HF-FD group exhibited increased circularity index and decreased ramification index, while MAC supplementation attenuated the alteration in these indices (Fig. 4i–k). Particularly, in the CA1 of HF-FD group, GFAP⁺ cells showed increased cell body hypertrophy and intense dendritic staining, which has previously reported in AD post-mortem tissue [43]. Importantly, MAC supplementation prevented the observed astroglial morphological profiles. Furthermore, MAC supplementation significantly prevented the upregulation of TNF-α, IL-1β, and IL-6 mRNA expression in the hippocampus induced by HF-FD diet (p < 0.05, Fig. 4f–h). These findings indicate that MACs prevented loss of the BBB’s tight junction proteins, gliosis, and neuroinflammation induced by HF-FD diet.

Following our observation on MACs reduce neuroinflammation induced by the HF-FD diet, we further evaluated PTP1B (a mediator cross-linking inflammation and abnormal synaptogenesis) and synaptic signaling pathway IRS-pAKT-pGSK3β-pTau, which plays an important role in synaptogenesis [44]. PTP1B levels in the hippocampus of HF-MAC group were significantly decreased compared with HF-FD group (p < 0.05) (Fig. 5a). Furthermore, MAC supplementation decreased the phosphorylation of IRS-1 at Ser307 (p < 0.05, Fig. 5b), which is associated with overexpression of PTP1B [45]. In addition, MAC supplementation prevented the downregulation of insulin signaling molecules that regulate synaptogenesis, including p-Akt Ser473 and p-GSK3β Ser9 induced by HF-FD diet (all p < 0.05, Fig. 5c, d). Furthermore, the level of p-Tau (S202 + T205) was decreased in the HF-MAC group compared with HF-FD group (p < 0.05, Fig. 5e).

Synaptogenesis signaling is important for normal synaptic ultrastructure and cognitive function. MAC supplementation significantly improved the synaptic ultrastructure with thicker postsynaptic densities and narrower synaptic clefts compared with the HF-FD group (Fig. 6a–c), examined by TEM. In line with this finding, presynaptic proteins, synapsin I (SYS) and synaptophysin (SYN), and postsynaptic protein, post-synaptic density 95 (PSD95) expression levels, were significantly increased in HF-MAC group compared with HF-FD group (all p < 0.05, Fig. 6d, e).

Considering our findings on MACs improved gut microbiota-brain axis, it is necessary to investigate if MAC supplement could prevent HF-FD diet-induced cognitive impairment. The nesting behavioral and temporal order memory tests were conducted to assess the ability to perform activities of daily living and recognition memory, which is impaired in AD patients [27, 46]. In the nesting behavioral test, the HF-MAC group had a higher deacon nest score and decreased untorn nestlet weight than that of HF-FD mice (p < 0.05) (Fig. 7a–c). Temporal order memory test analysis revealed MAC supplementation significantly increased the percentage of time spent with old familiar object (p < 0.05) (Fig. 7d and f) suggesting enhanced recognition memory, which was decreased by HF-FD diet. Total time spent in object exploration during the testing phase were comparable among the three groups (Fig. 7e). Therefore, declined cognitive function induced by HF-FD diet was prevented by the supplementation of MACs. Recent studies have demonstrated a clear association between changes in microbiota and cognitive behavior [47]. Pearson’s correlation analysis revealed a significant positive correlation between the abundance of Bacteroidetes and cognitive behavior, including the nest score in nesting behavioral tests (r = 0.77, p < 0.001) and discrimination ratio in temporal order memory test (r = 0.64, p = 0.005) (Fig. 7g and i) and a significant negative correlation between the abundance of Bacteroidetes and untorn nestlet weight (r = − 0.74, p < 0.001) (Fig. 7h). Furthermore, there was a significant negative correlation between the abundance of Proteobacteria and cognitive behavior, including nest score (r = − 0.81, p < 0.001) and discrimination ratio (r = − 0.54, p = 0.025) (Fig. S2A and B) and a significant positive correlation between the abundance of Proteobacteria and untorn nestlet weight (r = 0.80, p < 0.001) (Fig. S2C).

To investigate if gut microbiota is important for MAC effects on improving cognitive deficits induced by a HF-FD diet, a cocktail of oral antibiotics was used to eliminate the gut microbiota. Chronic oral antibiotic intervention reduced fecal bacterial loads in mice with MAC supplement in HF-FD diet (Fig. 8a). Furthermore, the effects of MAC supplement increasing colon length were attenuated by antibiotic intervention (Fig. 8b). Antibiotics alone did not significantly alter serum LPS levels in HF-FD group, suggesting antibiotics did not further increase endotoxin induced by HF-FD diet. However, serum LPS levels were significantly increased in HF-MAC group treated with antibiotic intervention compared with HF-MAC group (Fig. 8c). Compared with MAC supplement group, antibiotics significantly decreased deacon nest score and increased untorn nestlet weight in the nest behavior test (Fig. 8d, e) significantly attenuated recognition memory in temporal order memory test (Fig. 8f, g). Therefore, depletion of the gut microbiota using broad-spectrum antibiotics blocked MACs in improving cognition phenotype.

The datasets and/or analyzed during the current study are available from the corresponding author on reasonable request.