Dietary mixed-oxysterols and 27-Hydroxycholesterol induce cognitive impairment by regulating gut microbiota and miR-144-3p in vivo

Abstract

Gut microbiota and microRNAs (miRNAs) have been proved to be intimately involved in dementia. Our previous studies have showed that oxysterols and the subsequent neurotoxic effects contributed to the pathogenesis of cognitive decline. However, the exact mechanism linking dietary oxysterol-induced cognitive changes, gut microbiota, and miRNAs remains elusive. Here, two sets of experiments were conducted on male C57BL/6J mice treated with mixed-oxysterol diet or 27-hydroxycholesterol (27-OHC) combined with antibiotic cocktails and miRNA antagonists. Neurobehavioral tests were conducted to assess learning and memory of mice. 16S ribosomal DNA gene sequencing was performed to evaluate microbial diversity and community composition. Oxysterol levels were detected using HPLC–MS. Western blotting and RT-qPCR were used to detect the expression of the intestinal barrier-related factors. We found that a 0.05% mixed-oxysterol diet altered the gut microbiota, damaged the intestinal barrier, upregulated the expression of miR-144-3p, and resulted in learning and memory impairment, while depleting the gut microbiota with antibiotic cocktails partly alleviated these injuries. Moreover, there were enhanced Aβ deposition, as well as higher 27-OHC and its metabolite in the brain of oxysterols-treated mice, which could be reduced by sterol 27-hydroxylase inhibitor-anastrozole, indicating that 27-OHC might be the key regulator of oxysterol-induced brain pathological changes. Additionally, by antagonizing miR-144-3p, microbiota dysbiosis-related Aβ deposition, oxysterol load, and cognitive decline were significantly ameliorated. Taken together, our study demonstrates that dietary oxysterols impair cognitive function through 27-OHC causing microbiota dysbiosis and intestinal barrier dysfunction, targeting miR-144-3p might be a promising strategy against cognitive impairment.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-025-01628-3.

Introduction

Mild cognitive impairment (MCI) is defined as intermediate stage between the expected cognitive decline of normal aging and Alzheimer’s disease (AD) [1]. Over the past decades, despite the great efforts have been made in the study of cognitive concerns, the mechanisms of which have remained largely abstruse [2]. Given the continuous failure of therapeutic strategies targeting β-amyloid (Aβ) and tau protein in clinical trials, it is urgent to find early diagnostic biomarkers and explore the new intervention strategies for this complicated disease [3].

Recently, increasing evidence shows that the gut-microbiota-brain axis plays pivotal roles in regulating cognitive behaviours [4]. Human evidence shows that the gut microbiome diversity is decreased in AD individuals, as well as a decreased abundance in Firmicutes and Bifidobacterium but increased Bacteroidetes compared with the healthy controls [5]. Consistent with the results in AD patients, there are also alterations of gut microbiome in AD model mice, which have been proven to be linked to Aβ and tau pathology [6, 7] and neuroinflammation [4]. Besides, faecal microbiota transplantation from both APP/PS1 and C57BL6Cj wild-type donors into antibiotic-treated mice can restore Aβ amyloidosis and Aβ-associated degenerative changes [8], suggesting that the perturbation of gut microbiota is indeed responsible for the amyloidosis changes in AD.

As concluded by Lancet commission, 40% of dementia could be prevented or delayed by addressing modifiable risk factors, including healthy diet [9]. Changes in dietary patterns affect the composition and diversity of the gut microbiota [10, 11]. It has been reported that high cholesterol diet leads to a significant reduction in the diversity of intestinal flora in ApoE^−/− rats, and the microbiota features are similar to those found during the development of AD [12]. Recent studies have emphasized that oxysterols, especially 27-hydroxycholesterol (27-OHC), have great potential to reveal the effects of cholesterol metabolism dysfunction on cognitive function [13]. 27-OHC is an oxidative derivative of cholesterol hydroxylation by sterol 27-hydroxylase (CYP27A1) and is metabolized to 3β-hydroxy-5-cholestenoic acid (27-CA) and 7a-hydroxy-3-oxo-4-cholestenoic acid (7-HOCA) with the catalysis of CYP27A1, CYP7B1, and HSD3B7 [14], highlighting the essential regulation of CYP27A1 in cholesterol and oxysterol metabolism. Our previous study demonstrated that treatment with 27-OHC induced intestinal inflammation and altered the composition of the microbiome, leading to the aggravation of Aβ burden and cognitive deficits in APP/PS1 mice[15]. Nonetheless, whether dietary oxysterols affect cognition by disrupting gut microbiota is largely underexplored.

Dysregulated expression of microRNAs (miRNAs) is involved in several neurodegenerative disorders, including AD [16]. Further studies on the roles and mechanism of miRNAs in the pathological changes of brain will provide great insight into the AD progression and the use of miRNAs as potential biomarkers for AD. Our previous study has screened a variety of differentially expressed miRNAs in peripheral circulation of MCI individuals [17]. Among which, a significant increase in serum miR-144-3p but decrease in let-7g-5p, miR-107 and miR-186-3p were observed through our further verification test (the detailed data about the screening and verification of miRNAs are available in Supplementary Information 1) [18]. Recently, various studies have indicated that miRNAs are related to cognitive function by participating in cholesterol metabolism [19, 20]. For example, significant correlations between miR-144-3p and its target genes (3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) and ATP-binding cassette transporter A1 (ABCA1)) strengthen its roles in regulating cholesterol homeostasis [21]. Moreover, we also found that changes in miR‐144‐3p and let‐7g‐5p might be the molecular mechanism of 27‐OHC-induced abnormal Aβ metabolism in APP/PS1 mice [17]. These studies establish a strong association between miRNAs and oxysterol-driven cognitive decline, and the importance of studying miRNAs as therapeutic targets and biomarkers for gut dysbiosis is unveiled.

Based on the above evidence, we used mixed-oxysterol diet or 27-OHC injection combined with antibiotic cocktails to investigate the regulation of gut microbiota disturbance in oxysterol-induced cognitive decline. And miRNA antagomir was also used to reveal the possible mechanisms. Our study may provide support for regarding gut microbiota and miRNAs as intervention targets for cognitive impairment.

Material and methods

Animals and treatments

Male C57BL/6J mice aged 7 ~ 8 months were obtained from Beijing Vital River Laboratory Animal Technology Company and housed under SPF conditions at the Capital Medical University animal facility (30 ~ 35 g). The animals were raised under the condition of temperature 20 ~ 23 °C, humidity 50 ~ 55%, illumination 15 ~ 20 Lux, and a 12-h light–dark cycle. All the mice were allowed access to food and water ad libitum. The experiments were approved by the Ethics Committee of Capital Medical University (Ethics: AEEI-2018–188) and performed in accordance with the relevant ethical standards.

Before the formal experiment, we carried out a pre-experiment to find the dosage of mixed-oxysterol feeding in C57BL/6J mice to pave the road for further exploring and verifying the effects and mechanisms of mixed-oxysterol on cognitive function. According to the evidence of Brown et al., the mixed food was given at a dose of 4 g/kg total dietary sterols (0.4%), which corresponding to the amount found in human food. And it is estimated that approximately 12% of total sterols in mixed diet are oxysterols [22]. Accordingly, we calculated the dose of dietary oxysterols to be approximately 0.048% (≈0.05%). Based on the above evidence, we chose 0.05% as the median value, set up three dose groups around it with a ten-fold gradient (0.005%, 0.05%, 0.5%), and conducted a preliminary experiment. The results were shown in Supplementary Information 2. Overall, 0.05% was selected as effective dose for the formal experiment.

The formal experimental schedule is shown in Fig. 1. For the first set of animals, a total of 80 mice were randomly divided into eight groups as shown in Table 1 (n = 10/group). All animals were fed a standard chow and pure water for 1-week adaptation and then were given a 0.05% mixed-oxysterol diet or standard diet for 8 weeks, with or without antibiotic treatment, subcutaneous injection with anastrozole (ANS) 0.2 mg (1 mg/mL) or saline 0.2 mL every other day. The mixed-oxysterol diet was prepared as described previously [23]. In brief, pure cholesterol was heated at 150 °C for 12 h and added to the standard diet at a concentration of 0.05%. The formulas for the oxysterol and control diet are shown in Table 2. Antibiotic cocktails containing ampicillin (1 g/L), neomycin (1 g/L), metronidazole (1 g/L), and vancomycin (500 mg/L) were administered in drinking water to deplete the gut microbiota for 8 weeks [24].

Fig. 1.

Flow diagram of treatment and protocol design

Table 1.

Grouping and treatment of mice (part 1)

Number	Group	Treatment (n = 10)
1	CN	Control
2	AB	Antibiotic cocktails
3	ANS	CYP27A1 inhibitor-anastrozole
4	AA	ANS + antibiotic cocktails
5	OS	mixed-oxysterol diet
6	OAB	mixed-oxysterol diet + antibiotic cocktails
7	OAS	mixed-oxysterol diet + anastrozole
8	OAA	mixed-oxysterol diet + anastrozole + antibiotic cocktails

Table 2.

Formulas of the mixed-oxysterols diet and control diet

Ingredients (g)	0.05% mixed-oxysterol diet	Control diet
Casein	140	140
L-Cystine	1.8	1.8
Corn starch	495.7	495.7
Sucrose	100	100
Cellulose	50	50
Maltodextrin	125	125
Soybean oil	40	40
Mineral mix	35	35
Vitamin mix	10	10
Choline bitartrate	2.5	2.5
Oxysterols	0.5	0
Total	1000.5	1000

The second set of animals was randomly divided into seven groups as shown in Table 3 (n = 9/group). After 1-week adaptation, mice were subcutaneously injected with 27-OHC (5.5 mg/kg body weight) for 4 weeks, as described in our previous study [17]. A total dose of 5 nmol miRNA antagomir (antagomir-144-3p) or negative control was intranasally administered using a micropipette at a rate of 2 μL per nostril, with 2 ~ 3 min between each operation to allow the mice to breathe normally and inhale the administered drop. The treatment was performed on the first and 15th day of the experiment [25]. After the treatment, neurobehavioral tests were performed.

Table 3.

Grouping and treatment of mice (part 2)

Number	Group	Treatment (n = 9)
1	CN	Control
2	AB	Antibiotic cocktails
3	C27	27-OHC
4	C27AB	27-OHC + antibiotic cocktails
5	C27NC	27-OHC + miRNA negative control
6	C27AN	27-OHC + antibiotic cocktails + miRNA negative control
7	C27AA	27-OHC + antibiotic cocktails + antagomir-144-3p

Neurobehavioral tests

Novel object recognition test

The novel object recognition test was conducted in a square chamber (50 cm × 50 cm × 45 cm). Each mouse was placed in a test box for 5 min to adapt to the experimental environment. In the training phase 24 h later, two identical objects, A and B, were used as familiar objects and placed into the box at two corners with equal distance from the wall. Each mouse was allowed to explore the objects for 10 min. The test was conducted 24 h later, and object B was replaced with a novel object C. All animals were placed in the box from the same location, and the objects were explored for another 10 min. The object exploratory behaviour of the mice was defined as approaching (facing an object within 2 cm), sniffing, and touching the objects. The experimental field and objects were cleaned with 75% ethanol after each trial to eliminate odour cues. The exploration frequency and time of novel/familiar objects were recorded, and the novel object recognition index (NORI), frequency discrimination index (FDI), and time discrimination index (TDI) were calculated [26].

Morris water maze test (MWM)

The MWM test procedure was the same as our previous description [17]. Briefly, a circular steel pool (120 cm in diameter and 50 cm in height) was filled with opaque water and divided into four quadrants. The platform was submerged 1 cm below the water surface in the centre of the southeast quadrant, and its position remained constant. The high-contrast location cues were placed above the water surface and pasted on the walls of the pool. All mice were released into the pool from a predetermined position in each quadrant facing the wall. During the 5-day navigation tests, the maximum testing duration of each mouse was 90 s, and the time spent to reach the platform was recorded as the escape latency. The mice were guided to the platform and stayed for 15 s if they failed to find it within 90 s. Afterwards, the probe trials were conducted with the platform removed, and mice were released from the northwest quadrant and allowed to swim for 90 s. The results for each mouse were individually recorded.

Faecal microbiota profiling by 16S ribosomal DNA (rDNA) gene sequencing

16S rDNA sequencing was performed to characterize microbial diversity and community composition. Metagenomic DNA was extracted from faecal samples collected on the last day of treatment using a PowerFecal DNA Isolation Kit (MoBio, USA). The V3–V4 region of the 16S rDNA gene was amplified using barcoded primers. Polymerase chain reaction (PCR) was performed using High-Fidelity PCR Master Mix with GC Buffer (New England Biolabs, MA, USA) using the following procedure: pre-denaturation at 98 °C for 1 min, a total of 40 cycles including 98 ℃ for 10 s, 50 ℃ for 30 s, and 72 ℃ for 5 s and a final elongation of 72 °C for 5 min. Subsequently, the amplicons were purified, quantified, and sequenced using an Illumina NovaSeq6000 (Illumina, USA) [27]. QIIME (version 1.7.0) software was used to calculate the α-diversity indices including Shannon, Simpson, Chao1, and ACE. The β-diversity on both weighted and unweighted UniFrac was calculated by QIIME and visualized with principal coordinate analysis (PCoA). Linear discriminatory analysis effect size (LEfSe) was used to show the differences in bacterial taxa (P value < 0.05 by a Kruskal–Wallis test, linear discriminant analysis score > 3).

Quantification of oxysterols by HPLC–MS

Lipids were extracted from 50 mg brain tissue according to the modified Bligh and Dyer protocol [28]. A mixture of d7-24-hydroxcholesterol, d7-7β-hydroxycholesterol, d6-25-hydroxycholesterol, d7-7-keto-cholesterol, d7-7α-hydroxy-cholestenone, and d6-TMAS were used as internal standards. The reaction system was then subjected to derivatization. An Exion UPLC-QTRAP 6500 PLUS apparatus (Sciex, USA) equipped with electrospray ionization (ESI) was used for quantitative analysis.

Total RNA isolation and microRNA array assay

Total RNA from intestinal and brain tissues (18 ~ 20 mg) were isolated using an RNA Extraction Kit (TIANGEN Biotech Co., Ltd., China) according to the manufacturer’s instructions. RNA concentration was detected using a NanoDrop 2000 spectrophotometer and 1 μg of which was reverse-transcribed using the RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific, USA). The primers used are listed in Table 4. RT-qPCR was performed using the KAPA SYBR® PCR Kit (Kapa Biosystems, USA) on real-time qPCR detection system (Bio-Rad, USA). All samples were analysed in triplicate from at least three biological replicates.

Table 4.

Primers used in this study

Primer	Forward sequences (5′−3′)	Reverse sequences (5′−3′)
claudin7	GTCCCTGGTGTTGGGCTTCT	CGGGCCTTCTTCGCTTTGTC
mucin1	CAGTCTTGGCCACCACTCCA	GGTAGCACCGAGGAGCCATT
Abca1	AGAAGGAGGCTCGGCTGAAGG	GAGGGATGAGGCTGCTAACAAACC
Abcg1	CATGCTGCTGCCTCACCTCAC	TCTCGTCTGCCTTCATCCTTCTCC
Abca7	AGCCTCCTCTGTGGATGTCCTTG	GCTCTGGTGATGCGTTCCTCTATG
Cyp27a1	CACCGATGGCTGAGGAAGAAAGAG	ACCCAGGCAAGACCGAACCC
Hmgcr	GACCAACCTTCTACCTCAGCAAGC	CCAGCCATCACAGTGCCACATAC
β-actin	AGCGAGCATCCCCCAAAGTT	GGGCACGAAGGCTCATCATT

Western blot

Approximately 40 mg of intestinal or brain tissues were washed with PBS and lysed in freshly prepared RIPA buffer containing the protease inhibitor. The concentration of the protein extract was measured using the BCA Protein Quantitative Kit (Dingguo Changsheng Biotechnology, China). Protein samples (20 μg per lane) were separated using 10% sodium dodecyl sulphate–polyacrylamide gel electrophoresis and transferred onto polyvinylidene fluoride membranes (Merck Millipore Co., Ltd., Germany). The membranes were blocked with 5% skim milk at room temperature for 2 h, incubated with primary antibodies at 4 °C overnight, and incubated with HRP-conjugated secondary antibodies at room temperature for 1 h. Protein density was detected using Fusion FX (Vilber Lourmat, France) and normalized to β-actin levels.

Histopathological analysis

Immunohistochemical staining

Immunohistochemical staining for Aβ in the brain was performed using paraffin-embedded samples. The sections were subjected to dewaxing, antigen retrieval, blocking of endogenous peroxidase activity, and serum-sealing. Afterwards, the samples were incubated with the primary antibody overnight at 4 °C and the secondary antibody for 50 min at room temperature. Newly prepared DAB and haematoxylin staining solutions were added for colour development and nuclear counterstaining. The sections were then sealed with neutral gum and visualized under a microscope.

Haematoxylin–eosin (HE) staining

Whole brain, distal small intestine, and proximal colon tissues (~ 2 cm) were fixed in 4% paraformaldehyde and embedded in paraffin. The samples were cut into slides in a thickness of 5 μm and stained with haematoxylin and eosin dye for 5 min. After dehydration and sealing with neutral gum, slides were observed under a microscope.

Statistical analysis

SPSS 23.0 (Chicago, IL, USA) and GraphPad Prism 6.0 were applied for data analysis. For normally distributed data, the data are presented as mean ± standard error of the mean (SEM). One-way ANOVA was conducted to evaluate significant difference among groups, and post hoc comparisons were performed using the LSD-t OR Dunnett’s T3 test. A two-way ANOVA was performed for the repeated measurement analysis of the MWM test. P < 0.05 was considered statistically significant, and all tests were 2-sided.

Results

Dietary oxysterols disturbed the microbial composition and diversity

16S rDNA gene sequencing was conducted to investigate the community composition and diversity of the faecal microbiota. As shown in the flower diagram, 302 operational taxonomic units (OTUs) were displayed at 97% sequence similarity, and decreased OTUs numbers were obtained in the mixed-oxysterol diet treatment groups (OS, OAB, OAS, and OAA) (Fig. 2a). PCoA based on both weighted and unweighted UniFrac distance supported that the bacterial characteristics among the groups were significantly different (Fig. 2b).

Fig. 2.

Dietary oxysterols disturbed the microbial composition and diversity (n = 6 mice/group). a Flower diagram based on OTUs among all groups. b PCoA based on weighted and unweighted UniFrac distance. c The α-diversity of gut microbiota among groups (Shannon, Simpson, Chao1 and Ace). *P < 0.05; **P < 0.01. d The β-diversity of gut microbiota based on weighted and unweighted Unifrac among groups. e–f Sankey diagram for relative abundance of gut microbiota at phylum (e) and genus (f) level. g Linear discriminant analysis (LDA) scores by LEfSe > 3 and P < 0.05. h Taxonomic cladogram by LEfSe. Abbreviations: p, phylum; c, class; o, order; f, family; g, genus; s, species

The α-diversity (Fig. 2c) showed that compared with the CN group, the Shannon and Simpson indices of the AB, AA, OAB, and OAA groups were significantly decreased (P < 0.05). Meanwhile, decreased Chao1 and Ace indices in the AA, OS, OAS, and OAA groups demonstrated that treatment with dietary oxysterols and/or antibiotics contributed to the reduction in microbial abundance (P < 0.05). The β-diversity (Fig. 2d) results showed that the weighted UniFrac distances of OS (P_adj = 0.002) and AA (P_adj = 0.002) mice were significantly different from those of CN group. Additionally, noticeable changes were observed in the AB (P_adj = 0.004), AA (P_adj = 0.003), OS (P_adj = 0.030), OAB (P_adj = 0.002), and OAA (P_adj = 0.001) groups, based on the unweighted UniFrac distance.

The Sankey diagram revealed the relative abundance of the gut microbiota at different taxonomic levels. As illustrated in Fig. 2e, the most prominent species at the phylum level were Firmicutes, Bacteroidetes, Proteobacteria, and Verrucomicrobia. Treatment with antibiotics (AB) induced a higher abundance of Proteobacteria, but lower levels of Firmicutes and Bacteroidetes. Mixed-oxysterol diet (OS) increased the abundance of Firmicutes, Proteobacteria, and Verrucomicrobia but decreased the abundance of Bacteroidetes, while the addition of ANS (OAS) reduced the relative abundance of Proteobacteria. The top five species at the genus level were Enterobacter, Aeromonas, Lactobacillus, Escherichia, and Parabacteroides (Fig. 2f). Treatment with oxysterol diet increased the relative abundance of Enterobacter, but the abundance of which was decreased in the OAS group. By LEfSe analysis (Fig. 2g–h), a higher proportion of p_Proteobacteria was found in AB group; s_Bacteroide subgroups were more abundant in OS mice; the OAB group was mainly enriched with Enterobacteria at three levels, while ANS and CN were significantly enriched with Bacteroidetes.

Dietary oxysterols affected the intestinal structure and function through gut microbiota

HE staining of the ileum and colon was performed to evaluate the effects of the treatments on intestinal pathology. As illustrated in Fig. 3a–f, when compared to the CN mice, the ileac mucus layer thickness (P = 0.043) was remarkably shorter, while the crypt depth of colon (P < 0.001) was increased in OS group, and there was also a trend for the ileac villus height as well as the colonic mucus layer thickness to decrease. In comparison, ANS and AB treatment attenuated these histopathologic changes in mice.

Fig. 3.

Dietary oxysterols affected the intestinal structure and function through gut microbiota. a HE staining of ileum (scale bar, 20 μm; n = 3 mice/group). b Villus height of ileum. c Mucus layer thickness of ileum. d HE staining of colon (n = 3 mice/group; scale bar, 20 μm). e Crypt depth of colon. f Mucus layer thickness of colon. g–p The mRNA and protein expression of claudin7 and mucin1 in ileum (g–k) and colon (l–p) (n = 6 mice/group). g, i, l, n Claudin7; h, j, m, o mucin1; k, p western blot results. All data are presented as the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001

Simultaneously, the expression levels of claudin7 and mucin1 were measured to assess the intestinal barrier function. As shown in Fig. 3g–p, the expression levels of claudin7 and mucin1 in the OAB group were significantly lower than those in CN (claudin7 mRNA: P = 0.008; mucin1 mRNA: P = 0.008; mucin1 protein: P < 0.001) and AB (claudin7 mRNA: P < 0.001; mucin1 mRNA: P = 0.002; mucin1 protein: P = 0.003) mice in the ileum. Low levels of ileum mucin1 were also observed in the OS group compared with the CN (P = 0.023) and OAS (P = 0.046) groups at mRNA level. In the colon tissue, the expression of mucin1 mRNA was remarkably downregulated in the mixed-oxysterol treatment groups (OAB: P = 0.006; OAS: P = 0.007; OAA: P = 0.016) compared with the CN group; lower mucin1 levels were observed in the OAA group than in AA group (P < 0.001, P = 0.026). The expression level of mucin1 mRNA was significantly higher in AA mice than that in AB mice (P = 0.006), and the level of mucin1 mRNA in the ANS group was significantly higher than that in the OAS group (P = 0.002). The results suggested that dietary oxysterols significantly reduced the expression of intestinal barrier-related factors, while the administration of antibiotics and ANS distinctly alleviated these changes.

Dietary oxysterols affected the learning and memory through gut microbiota

Novel object recognition and MWM tests were conducted to assess the effects of oxysterols and gut microbiota on learning and memory in mice. In the novel object recognition test (Fig. 4a–c), compared with the CN group, the NORI, FDI, and TDI values of the AB (P = 0.032, P = 0.006, P = 0.004), AA (P = 0.009, P = 0.002, P = 0.026), and OS (P < 0.001, P < 0.001, P = 0.003) groups were significantly decreased, and the NORI and FDI in the OAB (P = 0.032, P = 0.042) and OAA (P = 0.025, P = 0.039) groups were also decreased. Moreover, compared with the OS group, the three indices in the OAS (P = 0.022, P < 0.001, P = 0.001) groups were evidently increased. The NORI and FDI values in the OAB (P = 0.042, P = 0.046) and OAA (P = 0.011, P = 0.011) groups were also higher than those in the OS mice.

Fig. 4.

Dietary oxysterols affected the learning and memory through gut microbiota (n = 10 mice/group). a–c Novel object recognition test. a Novel object recognition index. b Frequency discrimination index. c Time discrimination index. d–h Morris water mase test. d Escape latency. e Mean distance. f Average speed. g Number of crossing platforms. h Time in platform quadrant. All data are presented as the mean ± SEM. *P < 0.05; **P < 0.01. [Novel object recognition index (NORI) = (exploring frequency of novel object/total exploring frequency) × 100%, frequency discrimination index (FDI) = (exploring frequency of novel object − exploring frequency of familiar object)/(exploring frequency of novel object + exploring frequency of familiar object), time discrimination index (TDI) = (exploring time of novel object − exploring time of familiar object)/(exploring time of novel object + exploring time of familiar object)]

Regarding the MWM test, significant differences were observed in the intervention factors (F = 35.012, P < 0.001) as well as the interaction effects between the intervention factors and time (F = 7.991, P < 0.001). The escape latency to the platform during the 5-day training with marked differences is shown in Fig. 4d. (F = 542.491, P < 0.001). Mice treated with AB and OS showed longer escape latency (P < 0.001) than the CN group, which was shorter in ANS mice (P < 0.001). Compared to the OS group, the escape latency to platform was significantly decreased in the OAB (P = 0.011), OAS (P < 0.001), and OAA (P < 0.001) groups, while the latency in the OAB and OAA groups was significantly longer than that in the single treatment groups without dietary oxysterols (AB: P = 0.018, AA: P = 0.046). There was no significant difference in the mean distance to platform, average swimming speed, number of crossing platforms, and time spent in the platform quadrant (P > 0.05) (Fig. 4e–h). These results supported that the oxysterol treated mice performed worse in neurobehavioral tests, but with the using of antibiotics and ANS, mice were more capable of recognizing novel object and forming memories of the platform position through continuous training.

Dietary oxysterols affected the brain pathology through gut microbiota

HE staining was conducted to intuitively display the effects of treatments on brain pathology, and presented in Fig. 5a–b. OS treatment evidently aggravated the severity of morphological damage in the hippocampal CA1 region, as indicated by the reduced number and disordered arrangement of neurons, enlarged intercellular space, unclear cell membrane, and karyopyknosis. However, the treatment with OAB, OAS, and OAA mitigated the pathomorphological changes in mice.

Fig. 5.

Dietary oxysterols affected the brain pathology through gut microbiota a–b HE staining (a scale bar, 100 μm; b scale bar, 20 μm; n = 3 mice/group). c–d Immunohistochemical staining (c) and average optical density value (d) of Aβ in the brain (scale bar, 20 μm; n = 3 mice/group). e–g Relative expression of miR-144-3p (e), miR-186-3p (f), and let-7g-5p (g) in the brain (n = 6 mice/group). h–i The concentrations of 27-OHC (h) and 27-CA (i) in the brain (n = 4 mice/group). All data are presented as the mean ± SEM. *P < 0.05; **P < 0.01. Abbreviation: 27-OHC, 27-hydroxycholesterol; 27-CA, 3β-hydroxy-5-cholestenoic acid

Furthermore, the results of immunohistochemical staining (Fig. 5c) and calculation of average optical density (Fig. 5d) revealed that significantly increased Aβ deposition in the cerebral cortex were observed in the OS-treatment group compared to the CN group (OS, P < 0.001; OAB, P = 0.001; OAS, P = 0.001; OAA, P = 0.010). However, treatment with OS combined with antibiotics (OAB and OAA) resulted in an evident attenuation of Aβ deposition compared with OS mice (P < 0.001, P < 0.001).

Next, the miRNA expression levels in the brain (Fig. 5e–g) were measured using RT-qPCR. We noted that the expression level of miR-144-3p was significantly upregulated in the OS (P = 0.049) but downregulated in the OAA (P = 0.045) group compared with that in the CN mice. Moreover, its expression level was significantly lower in the OAB (P = 0.007), OAS (P < 0.001), and OAA (P < 0.001) groups than that in the OS group. Our data provided direct evidence that oxysterols could regulate the expression of miRNA in the brain, and miR-144-3p might be a potential target of dietary oxysterol-induced brain pathologic lesions.

Finally, the levels of 27-OHC and its metabolite (27-CA) were measured to assess brain oxysterol homeostasis after the treatments (Fig. 5h–i). We found that the 27-OHC (P = 0.019) and 27-CA (P = 0.014) levels were higher in OS group than in CN group. The 27-OHC level in the OAB group was significantly higher than that in the single antibiotic-treated (AB) group (P = 0.006). However, when compared to the OS group, the 27-OHC level was decreased in the OAS mice (P = 0.005), and the 27-CA level was decreased in OAB (P = 0.044) and OAA (P = 0.002) groups. Meanwhile, the levels of 27-OHC in the AA (P = 0.009) and 27-CA in the ANS (P = 0.033) groups were significantly lower than those in the CN group. These results supported the effects of dietary oxysterol on increasing 27-OHC and its metabolites in the brain and underlined the regulation of gut microbes and CYP27A1 on cerebral oxysterol homeostasis.

27-OHC affected brain Aβ and learning and memory by regulating the gut microbiota-related miR-144-3p expression

Having verifying the significant cognitive remission of dietary oxysterols combined with CYP27A1 inhibition, as well as the possible involvement of miR-144-3p, we further treated the mice with 27-OHC injection instead of oxysterol diet and simultaneously with a miR-144-3p antagonist to explore the potential molecular mechanism. As for the changes of brain histopathology, the area covered by Aβ plaques in the 27-OHC (C27 and C27NC) treated mice was significantly higher than that in the CN group (P < 0.001). However, mice in the C27AB and C27AA groups showed decreased plaque deposition compared with those in the C27 (P = 0.002) and C27AN (P = 0.032) groups (Fig. 6a–b).

Fig. 6.

27-OHC affects brain Aβ and learning and memory by regulating gut microbiota-related miR-144-3p expression (a–b). Immunohistochemical staining (a) and average optical density value (b) of Aβ in the brain (scale bar, 20 μm; n = 3 mice/group). c–e Novel object recognition test (n = 9 mice/group). c Novel object recognition index. d Frequency discrimination index. e Time discrimination index. f–j Morris water maze test (n = 9 mice/group). f Escape latency. g Mean distance. h Average speed. i Number of crossing platforms. j Time in platform quadrant. All data are presented as mean ± SEM. *P < 0.05; **P < 0.01. [The novel object recognition index (NORI) = (exploring frequency of novel object/total exploring frequency) × 100%, frequency discrimination index (FDI) = (exploring frequency of novel object – exploring frequency of familiar object)/(exploring frequency of novel object + exploring frequency of familiar object), and time discrimination index (TDI) = (exploring time of novel object – exploring time of familiar object) / (exploring time of novel object + exploring time of familiar object)]

In neurobehavioral tests, results from novel object recognition (Fig. 6c–e) showed that compared to CN group, the NORI (P < 0.001), FDI (P < 0.001), and TDI (P < 0.001) values were significantly decreased in the C27 mice. Nevertheless, the NORI and FDI values in the C27AB (NORI, P = 0.027; FDI, P = 0.033), C27AN (NORI, P = 0.028; FDI, P = 0.040), and C27AA (NORI, P = 0.044; FDI, P = 0.038) groups were significantly higher than those in the relative control (C27, C27NC, C27AN) groups, indicating that 27-OHC caused significant short-term/working memory loss in mice, but with a combination with antibiotics and antagomir-144-3p, the mice showed improved memory performance.

In the MWM test, significant differences were found in the place navigation test on day 5 (F = 8.105, P < 0.001). As illustrated in Fig. 6f, mice in C27 (P < 0.001), C27AB (P = 0.025), C27NC (P < 0.001), and C27AN (P = 0.021) groups displayed markedly longer escape latency to the platform. However, the latency to the platform of C27AN and C27AA groups was shorter than that of the C27NC and C27AN groups, respectively (P = 0.023, P = 0.028). No significant difference was observed in the spatial probe trial (P > 0.05) (Fig. 6g–j). These results confirmed the learning and memory deficits in 27-OHC treated mice. Likewise, using of antibiotic cocktails and antagomir-144-3p alleviated the cognitive changes induced by 27-OHC.

27-OHC affected the brain cholesterol and oxysterol homeostasis by regulating the gut microbiota-related miR-144-3p expression

The expression levels of factors involved in cholesterol synthesis, transport, and catabolism were measured to evaluate the alterations in cholesterol metabolism, which was the key precursor to 27-OHC synthesis (Fig. 7a–k). We found that the expression level of HMGCR and ABCA1 was significantly lower in C27 group than those in the CN group (HMGCR gene, P = 0.002; protein, P = 0.006; ABCA1 gene, P = 0.002; protein, P = 0.015) and C27AB (HMGCR gene, P = 0.007; protein, P = 0.003; ABCA1 gene, P = 0.021; protein, P = 0.002) group at both mRNA and protein levels. Meanwhile, significant upregulation of HMGCR (gene, P = 0.027) and ABCA1 (gene, P = 0.037; protein, P = 0.040) was observed in the C27AA treatment group compared with that in the C27AN mice.

Fig. 7.

27-OHC affects the brain cholesterol and oxysterol homeostasis by regulating gut microbiota-related miR-144-3p expression (a–k). The mRNA and protein expression of factors related to cholesterol metabolism in the brain (n = 6 mice/group). a, e HMGCR; b, f ABCA1; c, g ABCA7; d, h ABCG1; and i, j CYP27A1. k Western blot results. l Concentrations of oxysterols in the brain (n = 4 mice/group). All data are presented as mean ± SEM. *P < 0.05; **P < 0.01. HMGCR, hydroxymethyl glutaryl coenzyme A reductase; ABCA1, ATP-binding cassette transporter A1; ABCA7, ATP-binding cassette transporter A7; ABCG1, ATP-binding cassette transporter G1; CYP27A1, cytochrome P450 family 27 subfamily A polypeptide 1. 27-OHC, 27-hydroxycholesterol; 27-CA, 3β-hydroxy-5-cholestenoic acid; 7 K-27-OHC, 7-keto-27-hydroxycholesterol; 7-KC, 7-ketocholesterol; 7-OHC, 7-hydroxycholesterol; 4α-OHC, 4α-hydroxycholesterol; 4β-OHC, 4β-hydroxycholesterol; 24S-OHC, 24S-hydroxycholesterol; 25-OHC, 25-hydroxycholesterol; 7 K-25-OHC, 7-keto-25-hydroxycholesterol; 6 K-5α-OHC, 6-keto-5α-hydroxy cholesterol; 5α, 6α-EPOXY, 5α,6α-epoxycholesterol; 5β,6β-EPOXY, 5β,6β-epoxycholesterol; 3β,5α,6α-triol, Cholestane-3β,5α,6α-triol

As shown in Fig. 7i, there were significant differences in the total oxysterols (F = 6.510, P = 0.001), 27-OHC (F = 3.405, P = 0.017), 27-CA (F = 4.167, P = 0.007), 7 K-27-OHC (F = 7.741, P < 0.001), and 7-ketocholesterol (7-KC) (F = 3.958, P = 0.008) in the brain. Compared to the CN group, the levels of total oxysterols (P = 0.024), 27-OHC (P = 0.041), 27-CA (P = 0.026), and 7 K-27-OHC (P = 0.037) were remarkably increased in the C27 group. However, lower total oxysterols (P < 0.001), 27-CA (P < 0.001), and 7 K-27-OHC (P < 0.001) were observed in the C27AB group than in the C27 group. Moreover, the combined application of 27-OHC and antibiotic cocktails (C27AN) resulted in lower brain 27-OHC levels than that of C27NC group (P = 0.034). No significant difference was found in brain levels of 7-OHC, 4α-OHC, 4β-OHC, 24S-OHC, 25-OHC, 7 K-25-OHC, 6 K-5α-OHC, 5α, 6α-EPOXY, 5β, 6β-EPOXY, and 3β, 5α, 6α-triol.

Discussion

Emerging evidence has shown that oxysterols play important roles in neurodegenerative diseases, especially AD. However, whether dietary sources of oxysterols can alter cerebral oxysterol homeostasis and affect cognitive function remains unclear. In this study, we found that a 0.05% mixed-oxysterol diet altered the gut microbiota, induced intestinal barrier dysfunction, and upregulated the expression of miR-144-3p, leading to the learning and memory impairment. But using of CYP27A1 inhibitors and antibiotic cocktails alleviated these changes. 27-OHC-induced gut microbiota disturbance and the resulting cognitive decline could be improved by antagonizing miR-144-3p in the brain.

Several studies have highlighted the link between oxysterols and cognitive impairment [29, 30]. Our group also demonstrated that subcutaneous administration of oxysterols such as 27-OHC and 24S-OHC affected cognitive function through mechanisms of immunoregulation, neuroinflammation, etc. [15, 31]. However, recent researchers agree that dietary factors have an important impact on the risk of AD, promoting emerging interest in dietary oxysterols in this disease. Here, we found that mice fed a mixed-oxysterol diet showed significantly higher levels of 27-OHC and its metabolite 27-CA in the brain, exacerbating Aβ deposition in the cerebral cortex and cognitive impairment. Similar to our study, Khorrami et al. observed cognitive impairment in rats fed a 2% oxidized cholesterol-rich diet, the mechanism of which might be related to the decrease in antioxidant levels in the brain [32]. Liu et al. also suggested that dietary cholesterol oxidation products played harmful roles in neurodegenerative diseases through neurotoxicity, cytotoxicity, and pro-inflammatory pathways [33]. Optimizing the food processing and storage conditions, as well as adding antioxidants to restrict the production of cholesterol oxidation products, might be a preventive and therapeutic interventions to ameliorate its potential effects in various pathologies.

In addition, both α- and β-diversity analyses revealed a reduction in microbial diversity in dietary oxysterol treated mice, and evidence from 5 × FAD mice also supported that gut microbiota in AD tended to be lower than that in wild-type mice [34]. Analogous to our results, Zhang et al. reported that gut microbiota perturbation induced by both short- and long-term antibiotic treatment led to reduced Aβ deposition and alleviated plaque-localized glial reactivity in APP/PS1 mice [35]. Recently, the contribution of gut microbiota to neuropathology has been scientifically asserted [36, 37]. AD patients’ gut bacteria exacerbate AD-like pathologies in recipient mice, and the possible mechanisms include impairing hippocampal neurogenesis [38] and inducing the inflammatory pathway [4]. While faecal microbiota transplantation from wild-type mice into APP/PS1 mice significantly alleviated synaptic dysfunction, Aβ accumulation, neuroinflammation, and cognitive deficits, and the potential bacterial species responsible for gut dysbiosis might be strikingly decreased Firmicutes, Verrucomicrobia, Proteobacteria and Actinobacteria phyla, while increased Bacteroidetes and Tenericutes phyla [39, 40]. This is consistent with the gut microbiota signatures after dietary mixed-oxysterol treatment in our study.

The distinct changes of gut microbial profiles in AD should be the focus of further investigations. According to Zhang et al., the most frequently reported differential bacterial taxa are the phyla Proteobacteria and Verrucomicrobia; the family Lachnospiraceae, Enterococcaceae, and Ruminococcaceae; and the genera Bacteroides, Parabacteroides, Bifidobacterium, Dorea, Blautia, and Eubacterium. And for most of the reported differential microbial strains, contradictory results were observed among the studies [35]. Moreover, the microbiota profiles during AD progression are highly dynamic. Wang et al. found in 5 × FAD mice at the age of 2 months that the three most abundant bacteria at the phylum level were Bacteroides (47.3%), Firmicutes (33.0%), and Verrucomicrobia (12.2%), whereas at the age of 7 months, Firmicutes (62.8%) became the dominant phylum with a significant reduction in Verrucomicrobia and Bacteroidetes, indicating that the representative bacteria changed at different time point [2]. It is well-established that AD patients have a distinct gut microbial landscape compared to the healthy individuals, yet the substantial heterogeneity of which renders the conclusive results remain elusive. In addition, recent research has showed that gut microbes differ between the sexes, defined as microgenderome [41], and the findings of Jain et al. demonstrated that gut microbiota regulated neuroinflammation in a sex-specific manner in neurodegeneration [42], revealing a untapped intervention strategy for treating AD through the modulation of gut microbiota.

The cognitive deficits induced by oxysterols may be highly correlated with miR-144-3p in the brain according to our results, which sheds light on a new view of how oxysterols resulting in dysfunction of the gut-microbiota-brain axis. Research on miRNA dysregulation in AD pathogenesis has been widely conducted, and many miRNA-targeted genes have been shown to be directly involved in APP degradation and Aβ metabolism, inducing synaptic plasticity and the proliferation, migration, and differentiation of neurons [43, 44]. Evidence showed that compared with the down-regulation of global miRNA, miR-144 is the sole miRNA that consistently increased in the brains of elderly primate and AD patients, which displayed a significant negative correlation with cognitive performance [45, 46]. miR-144 induces the intracellular accumulation of reactive oxygen species (ROS) and reduces the activities of antioxidant enzymes [47], induces microglial autophagy and inflammation, and aggravates neuron degeneration and apoptosis [48], which partially reveals the possible mechanism of miR-144 participating in AD pathological process.

A recent in vitro study reported that miR-107-5p induced by 7-KC promoted differentiation and function of osteoclasts by downregulating mitogen-activated protein kinase[49]. Similarly, miR-7-5p as a 25-OHC-regulated miRNA could activate SREBP signalling and promote lipid accumulation in Huh7.5 hepatoma cells [50]. These findings support a possible link between oxysterols and miRNA in vitro, but very little evidence has verified the regulation of oxysterols on miRNAs in vivo, especially in the brain, and the mechanism of this association remains unclear. In this study, the administration of the miR-144-3p antagonist significantly downregulated its expression in the brain and alleviated the neuropathological effects of 27-OHC-intestinal dysbiosis signalling. In line with our study, study of Cheng et al. found that miR-144 silencing increased the expression and activity of CYP7B1, which was its direct target, and decreased the plasma level of 27-OHC, leading to enhancement of reverse cholesterol transport and oxysterol metabolism [51]. The above evidence established an association between oxysterols, especially 27-OHC, and miR-144-mediated signalling pathway.

Moreover, recent animal study of Lee et al. provided evidence for the interaction of miRNA-gut microbe in neurodegenerative diseases. According to their results, in Parkinson’s disease model mice, the expression level of miR-155-5p significantly upregulated in the brain and colon, and the alteration of the faecal microbiota was statistically correlated with the motor deficits [52]. Datta et al. found that Bacteroides fragilis had a strong potential to trigger the NF-kB-miR-146a-miR-155 signalling, conveying the gut microbiome-derived pathogenic signals into the brain [53]. Our previous human research has also found that in MCI participants, the class Gammaproteobacteria and phylum Proteobacteria were negatively correlated with hsa-miR-107 and hsa-let-7 g-5p, and the Shannon index were positively correlated with hsa-miR-186–3p and hsa-miR-107 [18]. Taken together, the current research could be an indication of the molecular mechanistic link between miR-144-3p and its contribution to gut-microbiota-brain axis dysfunction induced by oxysterols.

There are several limitations in this study. Though we have revealed the alterations of oxysterol homeostasis in the brain during gut microbiota dysbiosis, further studies to measure the oxysterol profiles in the gut and faeces will be critical to clarify its potential role in gut-microbiota-brain axis in AD pathology. Then, we conducted an 8-week feeding study with only male C57BL/6J mice, which complicated the generalization of our results from experimental animals to human beings. Given that the consumption of dietary cholesterol and oxysterols is long-term and chronic in real-life scenarios, it is quite necessary to extend the experimental duration. However, an increased period inevitably brings with it the cumulative influence of environmental factors [54], as well as the aging process of animals and the associated development of age-related pathologies [55–57]. Striking a balance between mitigating the potential drawbacks of long-term research and capturing the effects that beyond the experimental timeframe of the current intervention remains a significant challenge for our future study. In addition, the influence of gender on AD pathology is currently the focus of scientific research, particularly given the sex-specific association between gut microbiota and AD progression [42], further investigation will be required to apply AD model mice with both genders to enhance the universality and credibility of the results.

Conclusion

In summary, we have found that both oxysterol diet and 27-OHC injection induce gut microbiota dysbiosis and upregulate the expression of miR-144-3p, causing cerebral pathological injury and cognitive decline. But these changes can be mitigated by using antibiotic and antagonizing miR-144-3p. Our results also suggest that 27-OHC and CYP27A1 are the key factors regulating mixed oxysterol-induced pathological changes in the brain. Further studies on which specific category of gut microbiota serves as the primary taxa influencing neuropathology, as well as the molecular mechanism of miRNA, are needed. Maintaining the oxysterols homeostasis and antagonizing excessive miR-144-3p in the brain may represent a novel therapeutic strategy in gut-microbiota-brain research. Finally, it is necessary to validate our findings in clinically-relevant human populations.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

24S-OHC

24S-Hydroxycholesterol

25-OHC

25-Hydroxycholesterol

27-OHC

27-Hydroxycholesterol

27-CA

3β-Hydroxy-5-cholestenoic acid

7 K-27-OHC

7-Keto-27-hydroxycholesterol

7-KC

7-Ketocholesterol

ABCA1

ATP-binding cassette transporter A1

AD

Alzheimer’s disease

ANS

Anastrozole

Aβ

β-Amyloid

CYP27A1

Cytochrome P450 family 27 subfamily A polypeptide 1

CYP7B1

Cytochrome P450 family 7 subfamily B member 1

HMGCR

Hydroxymethyl glutaryl coenzyme A reductase

MCI

Mild cognitive impairment

miRNA

MicroRNA

Author contribution

R.X.: Conceptualization, funding acquisition, project administration, supervision, validation. T.W.: Data curation, formal analysis, investigation, methodology, visualization, writing—original draft. M.J.: Data curation, formal analysis, investigation, methodology, software, visualization, writing—original draft. X.Z.: Data curation, writing—review and editing. W.F.: Investigation, methodology. L.W.: Investigation, methodology. L.H.: Investigation, methodology. H.Y.: Investigation, methodology, project administration.

Funding

This work was supported by the State Key Program of the National Natural Science Foundation of China (grant number 81330065) and the National Natural Science Foundation of China (grant numbers 82173501 and 81673149).

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors agreed with the content and gave explicit consent to submit. The authors obtained consent from Capital Medical University before the work is submitted.

Conflict of interest

The authors declare no competing interests.