Cry2 deficiency leads to cognitive impairment through the microbiota-gut-brain axis mediated S1P/NLRP3/IL-1β pathway in mice

Abstract

Background

Alzheimer’s disease (AD) is characterized by extracellular Aβ deposition and tau hyperphosphorylation, leading to synaptic dysfunction and cognitive decline. Mounting evidence indicates that circadian rhythm disorders are associated with increased AD risks. Growing evidence implicates the microbiota-gut-brain axis and its metabolites as critical modulators of both circadian physiology and AD pathology. However, the molecular mechanism through which circadian disturbance modulates gut-brain communication to influence AD pathogenesis remains poorly understood.

Methods

Core circadian clock gene expression was assessed across four AD human brain datasets, and found Cry2 to be the only gene consistently downregulated. To investigate its functional role in vivo, we established a mouse model with hippocampal-specific Cry2 knockdown. Cognitive performance, gut microbiota composition, and metabolic alterations were evaluated using the Morris water maze, 16 S rRNA sequencing, and untargeted metabolomics, respectively. Intestinal barrier integrity, blood-brain barrier function, and neuroinflammatory signaling were examined through immunohistochemistry, immunofluorescence, and Western blotting. The contribution of microbiota disturbance was tested using fecal microbiota transplantation (FMT). The involvement of sphingolipid signaling was further assessed through FMT, pharmacological modulation with the S1PR agonist FTY720, NLRP3 knockout mice, and microglial assays.

Results

We found that the expression of Cry2 consistently decreased in the AD group in four AD-related datasets. Then, knockdown of Cry2 in the hippocampus (shCry2) caused dysbiosis of gut microbiota, intestinal barrier impairment, cognitive dysfunction and tau pathology in mice. Intriguingly, along with the disturbance in intestinal sphingolipid metabolism pathways, activation of the S1P/NLRP3/IL-1β pathway was found in the brain of shCry2 mice. Transplantation of “shCry2 microbiota” mimicked the pathological and behavioral changes induced by hippocampal Cry2 deficiency. Administration of S1PR agonist FTY720 significantly improved cognitive impairment and decreased the expression of NLRP3 in shCry2 mice, and knockdown of Cry2 in NLRP3^−/− mice alleviated tau pathology and cognitive impairment. FTY720 and S1PR1 antagonist W146 dose-dependently modulated the expression of NLRP3 in BV2 cells. Overexpressing Cry2 in the hippocampus significantly alleviated the tau pathology and cognitive decline in APP/PS1 mice.

Conclusion

Hippocampal Cry2 deficiency leads to cognitive impairment through the gut-brain axis mediated S1P/NLRP3/IL-1β pathway and might provide a potential therapeutic target for AD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12974-026-03706-5.

Background

Alzheimer’s disease (AD) is the most common dementia, which is characterized by Aβ plaque deposition and tau hyperphosphorylation, leading to neuronal and synaptic dysfunction, ultimately causing cognitive impairment [1, 2]. Epidemiological studies show that circadian rhythm sleep-wake disorders such as advanced or delayed sleep-wake phase disorder and non-24 h sleep-wake rhythm disorder are associated with AD risks [3, 4]. These circadian rhythms can be disturbed by environmental or genetic factors, leading to physiological and transcription changes. The circadian rhythm system, organized hierarchically in the suprachiasmatic nucleus (SCN), plays a crucial role in coordinating the internal biological clock of the organism to keep a 24-hour periodicity [5]. The central circadian clock is constituted of multiple oscillating neurons and clusters of astrocytes, which integrate into a circadian rhythm unit, producing signals to regulate the circadian cycle [6]. At the molecular level, the core clock is made up of an autoregulatory transcription-translation feedback loop (TTFL), the circadian locomotor output cycles kaput (Clock), and the brain and muscle Arnt-like protein 1 (Bmal1) in the cytoplasm activate the transcription of Period (Per1 and Per2) and Cryptochrome (Cry1 and Cry2) in mammals [7, 8]. Then the CRYs and PERs will enter the nucleus of cells and suppress the transcription activity of the BMAL1/CLOCK heterodimerization complex in a period of around 24 h [9, 10]. The circadian clock genes not only regulate the fundamental biological rhythm process but also have a significant impact on memory regulation in the hippocampus [11, 12]. APPswe/PS1dE9 transgenic mice show a reduced effect on the expression of Cry1 and Cry2, which is found to increase at night compared to day in wild-type control mice [13]. Peak expression of several clock-related genes (Nr1d1, Nr1d2, Per3, Cry1, Cry2, and Tef) is delayed in serum from older patients when compared to serum from young individuals [14]. However, the underlying molecular mechanisms by which circadian rhythm-related genes regulate AD pathology are yet to be fully explored.

Recently, it has been acknowledged that there exists a significant relationship between circadian rhythm disruption and the gut microbiota. It is indicated that the intestinal microbiota community of Clock-mutant mice showed lower taxonomic diversity compared to wild-type mice [15]. In monkeys, the ablation of BMAL1 results in an expansion of Bacteroidota at midnight and altered microbial oscillations, with disrupted rhythmicity of intestinal H₂O₂ derived from NOX1 being responsible for the changes in microbial oscillations [16]. Furthermore, emerging evidence suggests that gut microbiota also exhibit diurnal rhythms, manifested not only in the composition of microbial communities but also in the functional profiles [17]. Similarly, there is a bidirectional network between the gut microbiota and the central nervous system (CNS), which communicates through immune, endocrine/systemic, neural, metabolic pathways, and so on, influencing both physiological and cognitive functions [18]. Recent studies have indicated that the advancement and onset of AD are associated with dysbiosis of the gut microbiota and neuroinflammation [19, 20]. The gut microbiota from AD patients exacerbates the pathology of 3×Tg mice through the C/EBPβ/asparagine endopeptidase pathway [21]. Additionally, microbiome-produced metabolites can regulate the blood-brain barrier (BBB) and intestinal integrity, modulate mitochondrial network dynamics, and affect brain function [22, 23]. Together, lines of evidence remind us that gut microbiota is at the crossroads between circadian rhythms and AD [24].

Sphingolipids (SLs), including ceramide (N-acylated sphingosine), sphingosine, sphingomyelin, sphingosine kinase1 (Sphk1), and hundreds of different glycosphingolipids, are central in forming and maintaining cell membranes and contribute to cellular structure and function, participating in neural development and synaptic transmission pathways in the brain [25]. Imbalance in SLs metabolism has been implicated in neurodegenerative diseases, such as AD, Multiple Sclerosis, and depression [26, 27]. Previous studies had reported that a reduction in Sphk1 led to the dysfunction of inflammation resolution and defective microglial phagocytosis due to decreased secretion of specialized proresolving mediators (SPMs) [28]. The NLRP3/IL-1β pathway is a signaling cascade where activation of the NLRP3 inflammasome leads to the processing and release of pro-inflammatory cytokine IL-1β, contributing to inflammatory responses and host defense mechanisms [29, 30]. Evidence shows that the endogenous lipid metabolite sphingosine (Sph) acted as a damage-associated molecular pattern (DAMP) by inducing the NLRP3-inflammasome-dependent secretion of IL-1β from macrophages [31]. Additionally, sphingosine-1-phosphate (S1P) has been confirmed as a novel TREM2 ligand that promotes the phagocytosis of microglia [32]. Thus, it is hypothesized that alterations in sphingolipids in the brain may lead to changes in the downstream NLRP3/IL-1β signaling pathway. However, no direct and causal evidence among the altered gut microbiota, sphingolipid metabolism disturbance, and brain pathology has been demonstrated to date.

In the present study, we sought to explore the impact of decreased hippocampal expression of Cry2 on cognitive function. We found that decreased expression of Cry2 in the hippocampus could induce changes in the composition and diversity of the gut microbiota and reduce the production of intestinal metabolites such as sphingolipids in the gut, resulting in intestinal barrier integrity and BBB impairment. This led to activation of the S1P/NLRP3/IL-1β pathway in the brain, ultimately resulting in cognitive dysfunction and increased tau pathology. A comprehensive understanding of the molecular mechanism by which Cry2 exerted its function in the progression of AD could provide significant support for future clinical studies aimed at assessing the therapeutic potential of Cry2 in AD.

Methods

Animals and intraperitoneal (i.p.) administration

The C57BL/6J mice were purchased from the Model Animal Research Center of Nanjing University (Nanjing, China). NLRP3^−/− (B6.129S6-Nlrp3^tm1Bhk/J, strain# 021302, RRID: IMSR_JAX:021302) and APP/PS1 mice (B6; C3-Tg(APPswe, PSEN1dE9)85Dbo/Mmjax, strain# 034829-JAX, RRID: MMRRC_034829-JAX) were obtained from Jackson Laboratory (Shanghai, China). They were all raised in the Experimental Animal Center of Southeast University School of Medicine under the condition of specific-pathogen-free conditions (SPF), maintaining a suitable temperature (22 ± 2℃) and humidity (50 ± 5%). The mice were housed under a regular 12:12 light-dark cycle and were provided with water and food ad libitum. For drug administration, FTY720 was diluted with sterile saline and administered via intraperitoneal injections at a concentration of 1 mg/kg once daily for two weeks. All animal-related procedures were performed in accordance with ethical standards and were approved by the Institutional Animal Care and Use Committee of Southeast University (20210924075).

Cell culture

Mouse microglia BV2 cell line was cultured in DMEM basic medium supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin antibiotics, and 1% MycoCleaner Reagent at 37℃ in a 5% CO₂ incubator. Lipopolysaccharide (LPS, L4516) derived from Escherichia coli serotype O127:B8 was purchased from Sigma-Aldrich (St. Louis, MO, USA). FTY720 (HY-12005) and W146 (HY-101395) were purchased from MedChemExpress (NJ, USA). The BV2 cells were pre-treated with LPS (5ng/ml) for 30 min, followed by treatment with different concentrations of FTY720 (1µM, 2.5µM, 5µM, 10µM), and W146 (1µM, 10µM) for 24 h.

Adeno-Associated Virus (AAV) construction and stereotaxic injection administration

All AAVs were designed by GeneChem Co., Ltd (Shanghai, China). The sequences for vectors harboring short hairpin RNA (shRNA) sequences targeting Cry2 (shCry2) and a scrambled sequence control (shScramble, hereafter abbreviated as shScr in figures) were GGATGAATGCCAATTCCTTAC and CGCTGAGTACTTCGAAATGTC, respectively. The overexpressing AAVs, AAV-Cry2-GFP and AAV-control-GFP vectors were constructed from AAV serotype 9. All AAVs carried a GFP tag for subsequent experimental validation. After inducing anesthesia with 4% pentobarbital (i.p. injection), mice were secured in a stereotaxic apparatus (David Kopf Instruments, Tujunga, Calif., USA). A 1 mm diameter hole was drilled at the coordinates corresponding to the exposed skull (ML = ± 2.0 mm, AP = − 2.0 mm, DV = − 2.0 mm). Subsequently, 1.0 µl AAV vectors were injected into each side of the brain of the mice at a speed of 0.2 µl/min to access the hippocampus. The microsyringe was left in place for 10 min following injection to minimize the backflow of injected viral particles.

Morris Water Maze (MWM) test

The MWM was used to assess spatial learning and memory function in mice. Procedures were conducted according to the previously published protocol [33]. The circular container, with a 120 cm diameter and a height of 40 cm, was filled with water made opaque by using Edible white pigment. The container was divided into four quadrants (SW, NE, NW, and SE), with a submerged circular escape platform (10 cm in diameter) placed in the SE quadrant. The temperature of the water was consistently maintained at 22 ± 2℃, and consistent lighting and visual cues were ensured around the maze. The MWM tests were conducted at ZT 6 (8:00 a.m. as ZT0). Throughout the training phase, mice were placed into the water from four different quadrants, training them to find the platform within 60s. If the mice successfully located the platform, they were allowed to stay on it for 15 s; if the mice failed to find the platform, they were gently guided to stay on the platform for a duration of 30 s. During the probe trial, the mice were allowed to swim freely for 60 s after removing the platform on the sixth day to assess memory retention. The swimming paths and the escape latencies of the mice were monitored and analyzed using Any-Maze software (Stoelting Co., Wood Dale, IL, USA). The recorded metrics included the time spent in the target quadrant and the number of platform crossings. The data were then analyzed with GraphPad Prism 9.0 (San Diego, CA, USA).

Tissue collection

The day after the MWM, we conducted tissue sampling on the mice. Mice were administered deep anesthesia with 4% pentobarbital (i.p. injection). Then the abdominal cavity was opened to expose the heart, and cardiac perfusion was performed with 0.1 M PBS on the mice. If immunostaining or immunofluorescence of the tissue was required, an additional cardiac perfusion step with 4% polyformaldehyde was performed after the initial PBS perfusion. Then the brain and colon tissues were prepared for further analysis.

Western blot

The procedures of the Western blot were performed following a previously established protocol [34]. Hippocampus and colon tissues were lysed in RIPA lysis buffer containing phosphatase inhibitor and protease inhibitor (1%) on ice for 30 min. The homogenates were centrifuged at 12,000 g at 4℃ for 15 min. After collecting the supernatant, protein concentration was determined using the Bicinchoninic Acid (BCA) Assay Kit (KeyGEN BioTECH, Nanjing, China). Then the protein was diluted with 5× loading buffer and boiled at 100 °C for 6 min. A total of 15–20 µg of protein (per lane) was loaded on the 10% SDS-PAGE gels and subsequently transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, USA). The membranes were blocked by 1×TBST containing 5% skim milk for 1 h at room temperature. Then the membranes were incubated overnight at 4 ℃ with the primary antibodies [Tau5 (1:1000, ab80579, Abcam), Tau1 (1:1000, MAB3420, MilliporeSigma), pS396-tau (1:5000, ab109390, Abcam), pT231-tau (1:5000, ab151559, Abcam), pT181-tau (1:5000, ab254409, Abcam), pT217-tau (1:2000, 44–744, Invitrogen), Occludin (1:1000, ab216327, Abcam), ZO-1 (1:1000, ab307799, Abcam), Sphk1 (1:1000, 10670-1-AP, Proteintech), CRY2 (1:1000, 13997-1-AP, Proteintech), S1PR1 (1:1000, 55133-1-AP, Proteintech), β-actin (1:5000, 66009-1-Ig, Proteintech), NLRP3 (1:1000, ab263899, Abcam), IL-1β (1:1000, ab234437, Abcam)] diluted in 1×TBST and washed 10 min three times. The membranes were incubated with horseradish peroxidase-conjugated (HRP)-labeled secondary antibodies at room temperature for 1 h. Finally, the membranes were visualized using the enhanced chemiluminescence substrate (ECL, Millipore, USA), and the Image Pro-Plus 6.0 software was employed to quantify the integral optical density values (IOD). Protein expression was analyzed using a relative quantification method, with the control group normalized to 1.

Immunostaining

The colon tissues were longitudinally opened along the mesenteric edge, and the fecal contents were carefully removed. Then tissues were rolled from the proximal to the distal, adopting a Swiss-roll-like configuration. The colons were fixed with Carnoy’s Fluid to perform the immunohistochemical assay. The colon and brains were fixed in 4% paraformaldehyde for immunofluorescence. Then the colons and brains underwent dehydration in progressively increasing concentrations of ethanol(70%, 80%, 95%, 100%, 100%), followed by clearing with xylene to remove residual ethanol. The specimens were embedded in liquid paraffin and sliced into sections of 5 μm for subsequent staining. we performed Hematoxylin and Eosin (HE) staining as well as Alcian Blue staining on the colon tissues for Glands, Epithelial Tissues, and connective tissues of the colons.

For immunofluorescence of brain tissues, the sections underwent deparaffinization and rehydration, followed by antigen retrieval. Next, the tissues were blocked with the 3% bovine serum albumin (BSA) at room temperature for 30 min. This was followed by incubation at 4℃ overnight in the humid box with antibodies targeting pS396-tau (1:200, ab109390, Abcam), ZO-1 (1:200, ab307799, Abcam), Iba-1 (1:200, ab178846, Abcam), CRY2 (1:50, 13997-1-AP, Proteintech) and MUC2 (1:200, ab272692, Abcam). Then the sections were incubated with an Alexa Fluor secondary antibody (Beyotime, Shanghai, China) for 2 h at room temperature, followed by a coverslipping with an Antifade Mounting Medium containing DAPI. All sections were observed under a confocal microscope (ZEISS, Germany) and analyzed using the ImageJ software.

Bioinformatics analysis

The gene expression datasets (GSE138260 [35], GSE29378 [36], GSE48350 [37], GSE5281 [38]) were obtained from the Gene Expression Omnibus (GEO) repository. Microarray annotation information from GEO datasets on NCBI was used to match probes with their corresponding genes. All datasets were standardized using the limma package, and the analysis was performed using R software [39, 40].

Antibiotic treatment and Fecal Microbiota Transfer (FMT)

Following the previously published protocol [41], male C57BL/6J mice aged 6–8 weeks were subjected to a 7-week regimen of antibiotic administration through drinking water. The antibiotic mixture included vancomycin (500 mg/L, Macklin, Shanghai, China), imipenem plus cilastatin (250 mg/L, MSD, Kenilworth, NJ, USA), ampicillin (1 g/L, Meryer, Shanghai, China), ciprofloxacin (200 mg/L, Macklin, Shanghai, China), metronidazole (1 g/L, Aladdin, Shanghai, China). The antibiotic water bottles were inverted daily, and the antibiotic solutions were renewed every 2–3 days. Subsequently, 72 h prior to the reestablishment of gut microbiota, the antibiotic treatment was discontinued, and sterile tap water replaced it.

According to the established protocol [34], the procedures of FMT could be summarized as follows. Briefly, mice were administered an oral gavage of 200 µl of fecal dilution for 7 consecutive days. The fecal dilution was prepared using 20 ~ 30 mg of fresh stool collected from shScramble and shCry2 donor mice, respectively. After pooling the fecal samples, they were dissolved in PBS at a ratio of 20 mg/100 µl until almost completely diluted. The samples were centrifuged at 50 g for 2 min to collect the supernatant, which was then diluted 1:1 with sterile PBS. Then the mice were orally gavage administered with the suspension.

Fecal sample collection and 16 S rRNA gene sequencing analysis

Mice were placed in a clean chamber with fresh water and food to collect fecal samples after 6 weeks of AAV stereotaxic injection. For each mouse, 3–5 fecal pellets were placed in clean, dry 1.5 ml centrifuge tubes and stored at -80℃ for 16 S rRNA sequencing and metabonomics analysis. All fecal pellets were obtained at ZT8. Total genomic DNA from the stool samples was extracted and diluted to 1ng/µl with sterile water. The V3-V4 variable regions of the 16 S rRNA gene were amplified with the standard of samples in agarose gel with a bright main strip between 400 and 450 bp for further experiments. The PCR products were mixed and purified with an AxyPrepDNA Gel Extraction Kit (Axygen, USA) and then sequenced on an Illumina NovaSeq 6000 platform. The data was then analyzed by QIIME2 according to the official tutorial (https://github.com/qiime2). Briefly, the sequencing data were demultiplexed, primer-trimmed, quality-filtered, denoised, and merged with the plugins in QIIME2. Based on the denoising and clustering results, species annotation was performed with amplicon sequence variants (ASV) for each sequence to obtain corresponding species information and the distribution of species abundance. Alpha-diversity metrics(Chao1, Observed Species, Simpson, Ace, and Shannon index) and beta-diversity metrics(weighted UniFrac, Bray-Curtis dissimilarity, and unweighted UniFrac) were calculated using the diversityplugin. Unweighted unifrac distance was used for Principal Coordinate Analysis (PCoA). STAMP software was employed to confirm differences in the abundances of individual taxonomy between the two groups, and LEfSe was utilized for the quantitative analysis of biomarkers within different groups.

Metabolomics analysis

The Metabolomics of the fecal stool samples were measured by Applied Protein Technology, Inc., Shanghai, China, with an untargeted LC/MS platform. Then, the Kyoto Encyclopedia of Genes and Genomes (KEGG) database calculated the signaling pathways of differential metabolites by matching the IDs of differential metabolites with pathway information in the KEGG database [42]. Fisher’s exact test was used to identify pathways affected by differential metabolites, considering the entire set of metabolites within each pathway as the background dataset. Pathways with p-values < 0.05 were considered as significantly changed pathways. Integrated analysis of the differentially expressed microbial communities and metabolites identified in samples was analyzed using the Spearman Correlation Coefficient. This approach aimed to elucidate the interaction relationships between microbial communities and metabolites.

Lipidomics analysis

The bilateral hippocampus of the mice was transferred to 1.5 ml centrifuge tubes and mixed with 200µL of pre-ice-cold water. After slow thawing at 4 °C, 1mL of pre-ice-cold methanol/acetonitrile/water (2:2:1) was added. The mixture was vortexed thoroughly, followed by cold ultrasonication for 30 min (pulse on 5s, pulse off 5s, amplitude 30%). Samples were then incubated at -20 °C for 10 min and centrifuged at 14,000 g for 20 min at 4 °C. The supernatant was collected and dried under vacuum. For mass spectrometric analysis, the dried residue was reconstituted in 100 µL of acetonitrile/water (1:1), vortexed, and centrifuged again at 14,000 g for 15 min at 4 °C. The supernatant was then transferred to autosampler vials for LC-MS analysis. For hippocampal analysis, 75 µL of supernatant was analyzed by LC/MS analysis using the UHPLC system (1290, Agilent Technologies) equipped with a Kinetex C18 column for targeted lipidomic. Every group had 10 replicates. The lipidomic data analysis was performed using the CentWave algorithm in XCMS for peak detection, extraction, alignment, and integration. Lipid identification was conducted through a spectral match using the lipidIMMS library, developed using R based on XCMS and lipid4danalyzer software.

Statistical analysis

All data were presented as mean ± standard error of the mean (SEM), and statistical analyses were performed using GraphPad Prism 9.0 software (GraphPad Inc., CA, USA). Statistical significance was assessed by unpaired Student’s t-tests for two independent groups or one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test for multiple comparisons. However, if the data did not follow a normal distribution (as determined by the D’Agostino-Pearson omnibus normality test) or if variances were unequal, the Kruskal-Wallis test or Mann-Whitney test was used instead. A repeated measures two-way ANOVA was used to compare the data from different groups at different time points, which evaluated changes within animals during MWM spatial training over time. p values < 0.05 were considered statistically significant.

Results

Effect of hippocampal Cry2 knockdown on the cognitive function and tau pathology in mice

For better understanding the role of circadian rhythm disorder in AD, we compared the expression of common clock genes, including PER1, PER2, PER3, CRY1, CRY2, ARNTL (BMAL1), and CLOCK in four AD-related datasets. These four datasets are derived from the hippocampal tissues of AD patients and their respective control groups. We found that only the expression of Cry2 consistently decreased in the hippocampus of the AD group compared to the control group (Fig. 1A, Supplementary Fig. 1A-E). CRY2 protein expression was validated in 10-month-old APP/PS1 mice compared to age-matched WT controls (Fig. 1B). Sought to investigate whether Cry2 regulated cognitive function, we specifically decreased Cry2 expression levels in the hippocampus using adeno-associated virus (AAV) vectors (Fig. 1C) and reduced expression of CRY2 was confirmed by Western blot and immunofluorescence assay in the hippocampus of C57BL/6 mice (Fig. 1D-E, Supplementary Fig. 5A). In the MWM test, across the five training days, a prolonged escape latency was consistently observed in the shCry2 group compared to the shScramble group, particularly notable on the fourth and fifth days (Fig. 1F). In the spatial probe test, the number of crossing the platform and the time spent in the platform quadrant in shCry2 was obviously lower than shScramble group (Fig. 1G-H). The swimming heatmaps of the mice in the spatial probe test are shown in Fig. 1I. To determine whether Cry2 knockdown could contribute to AD pathology, we evaluated the protein expression of tau-related pathology. The protein expression levels of pS396-tau, pT231-tau, pT181-tau, and pT217-tau were significantly elevated in the shCry2 group compared to the shScramble group, whereas the expression of Tau1 was decreased, and no significant difference was observed in the expression of Tau5 (Fig. 1J-K). Additionally, immunofluorescence results confirmed an increase in pS396-tau expression in the hippocampus of the shCry2 group when compared to the shScramble group (Fig. 1L-M). These findings indicated that Cry2 deficiency promoted the development of tau pathology and cognitive decline in mice.

Fig. 1.

Cry2 deficiency induces tau pathology and cognitive decline. (A) CRY2 gene expression levels in four AD datasets., (B) Representative Western blot and quantification of CRY2 in the hippocampus of 10-month-old APP/PS1 mice compared to their littermates., (C) An overview of the experimental design. Three-month-old mice received a stereotaxic injection, and the behavioral tests were conducted six weeks later, followed by a series of pathological and molecular biology tests., (D) Representative images of shCry2-AAV GFP fluorescence, (green) expression levels in the hippocampus of the brain. Magnification × 10. Scale bar = 100 μm., (E) Representative Western blot and quantification of CRY2 in the hippocampus lysates of shScramble, (hereafter abbreviated as shScr in the figures) and shCry2., (F-I) Escape latency of shScr and shCry2 for 5 consecutive days in the MWM test, (F); Percentage of time spent in the target quadrant in the spatial probe trail of the MWM test, (H); Number of platform crossings in the spatial probe trail of the MWM test, (G); Representative heatmaps of the swimming path in the spatial probe trail of day 6 in the MWM test, (I), (n = 12 mice/group)., (J-K) Representative Western blot and quantification of Tau1, Tau5, pS396-tau, pT181-tau, pT217-tau, and pT231-tau in the hippocampus of shScr and shCry2., (L-M) Representative immunofluorescence staining and quantification of pS396-tau, (red) in the hippocampus of shScr and shCry2 mice. Magnification × 10. Scale bar = 100 μm. All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using a repeated measures 2-way ANOVA test with Bonferroni post hoc comparison in panel, (F), and the Student’s t-test in panel, (A, B, E, G, H, J, M). MWM: the Morris Water Maze; AD: Alzheimer’s disease; FI: Fluorescence intensity. shScr: shScramble

Effect of hippocampal Cry2 knockdown on the composition of gut microbiota in mice

To determine whether the observed cognitive decline in shCry2 mice was due to altered gut microbiota, 16 S rRNA sequencing was conducted to detect the bacterial taxonomic composition. A total of 18 samples were collected from two groups (n = 9) of mice and then sequenced to generate 16 S rRNA gene profiles. Although the shCry2 group showed no significant changes when compared to the shScramble group in the Alpha-Diversity, as indicated by Simpson. Shannon, Ace, Chao1, and Observed species, there was an increasing trend between the two groups (Fig. 2A). The Beta-Diversity was measured by calculating the principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity distance matrices between the two groups. A clear separation was indicated between the shCry2 and shScramble groups (Fig. 2E). We further analyzed the variant taxa that occupied high abundance at different taxonomic levels. The results indicated that at the phylum level, Bacteroidetes, Actinobacteria, and Proteobacteria were significantly reduced in abundance, while Firmicutes and Desulfobacterota were more abundant in the shCry2 group compared with the shScramble group (Fig. 2B). At the family level, shCry2 were associated with a decreased abundance of Muribaculaceae and Akkermansiaceae, and an increased relative abundance of Lachnospiraceae, Oscillospiraceae and Ruminococcaceae (Fig. 2C). Additionally, at the level of genus, a significant increase in Lachnospiraceae NK4A136 and a decrease in the abundance of Muribaculaceae, Dubosiella, and Akkermansia were observed in the shCry2 group compared to the shScramble group (Fig. 2D). At the species level, the shCry2 group showed increased expression of Clostridium sp, Lachnospiraceae bacterium, and Streptococcus danieliae, and decreased expression of Firmicutes bacterium, Bacteroides caecimuris, Christensenella minuta, and Parabacteroides gordonii compared to the shScramble group (Supplementary Fig. 2C). LefSe (linear discriminant analysis Effect Size) was used to identify the specific microbiota taxa that were enriched in the shScramble and shCry2 groups separately. It not only selected microbial taxa that had significant differences in abundance but also had a large effect size. A representative cladogram depicted the distribution of bacterial taxa with different abundance in the shCry2 group and the shScramble group (Supplementary Fig. 2A). Based on the LDA (Linear Discriminant Analysis) scores, the biomarkers are significantly different between the shScramble and shCry2 groups (Supplementary Fig. 2B). These findings suggested that the detrimental effects of Cry2 on cognitive function might be mediated in large part by dysbiosis of intestinal microbiota.

Fig. 2.

Gut microbiota composition is altered in shCry2 mice. (A) Relative abundance of microbiota alpha diversity, including Observed species, Shannon index, Simpson index, Ace index, and Chao1 index in the shCry2 and shScr mice, (n = 9 mice/group)., (B-D) Relative abundances of phylum-level, (B), family-level, (C), and genus-level, (D) of gut microbial taxa in the shCry2 and the shScr mice. Each box represents the 25th to 75th percentiles, (n = 9 mice/group)., (E) Microbial communities clustered using PCoA of the Bray-Curtis Distance. The percentage of variation explained by the principal coordinates was indicated on the axes, (n = 9 mice/group). All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using the Mann-Whitney test., (A-D). shScr: shScramble

Effect of hippocampal Cry2 knockdown on intestinal barrier and BBB in mice

Recent studies indicated that the bidirectional relationship between gut microbiota and brain pathology was established through compromised intestinal and blood-brain barriers [43]. To determine the integrity of the intestinal barrier and BBB, ZO-1 and Occludin were measured by Western blot. ZO-1 and Occludin were both members of the tight junction protein family, played crucial roles in the cell-to-cell junction of vascular endothelial cells, and their decreased expression indicated impaired barrier function. Compared to the shScramble group, the shCry2 group showed a decreased protein expression of ZO-1 and Occludin in both the hippocampus and colon (Fig. 3A, C). Furthermore, in H&E-stained colon sections, inflammatory cell infiltration was observed in the intestinal epithelial mucosa and submucosa, accompanied by a decrease in the number of goblet cells, reduced height of colonic villi, and increased depth of crypts in the shCry2 group compared to the shScramble group (Fig. 3B, E-G). Additionally, Alcian blue staining revealed a thinner and discontinuous mucus layer, along with a loss of mucin-secreting goblet cells in the shCry2 group compared to the shScramble group (Fig. 3B, E). Mucin-2 (MUC2) is a mucin protein secreted by goblet cells, vital for protecting the intestinal mucosa by maintaining the stability of the mucus layer, serving as an indicator of intestinal barrier function. Immunofluorescence staining of ZO-1 and MUC2 illustrated the impairment of the intestinal mucus layer in the shCry2 group compared to the shScramble group (Fig. 3B, F-G). The shCry2 group had significantly reduced expression of ZO-1 in the hippocampus compared to the shScramble group by immunofluorescence staining (Fig. 3D, H). Taken together, these data indicated that hippocampal Cry2 deficiency led to the impairment of the intestinal barrier and the blood-brain barrier.

Fig. 3.

Cry2 deficiency impairs the intestinal barrier and blood-brain barrier. (A) Representative Western blot images and quantification of Occludin and ZO-1 in the colon of shScr and shCry2 mice. (B) H&E staining (left), Alcian blue staining (left), and representative ZO-1 immunofluorescence staining (right) were performed on paraffin-embedded colon sections. Representative immunofluorescence staining (right) against mucin-2 was performed on Carnoy-fixed colon sections. Magnification × 20. Scale bar = 100 μm. (C) Representative Western blot images and quantification of Occludin and ZO-1 in the hippocampus of shScr and shCry2 mice. (D, H) Representative immunofluorescence staining and quantification of ZO-1 (red) in the hippocampus of shScr and shCry2 mice. Magnification × 10. Scale bar = 100 μm. (E-G) Quantification of (B). All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using the Student’s t-test in panel (A, C, E-H). AB: Alcian blue; FI: Fluorescence intensity. shScr: shScramble

Transfer of “shCry2 microbiota” induced cognitive impairment and tau-related pathology

To further investigate whether the cognitive decline, memory loss, and tau-related pathology induced by hippocampal Cry2 deficiency depended on the gut microbiota, fecal microbiota was transplanted from shScramble and shCry2 group mice into gut microbiota-loss mice (which underwent antibiotics pretreatment) for 7 days (Fig. 4A). After one week of FMT, the feces from the donor and recipient mice were collected for 16 S rRNA gene sequencing to confirm whether the clustering variations in microbiota community structure between shScramble and shCry2 group mice were maintained in the FMT/shScramble group and FMT/shCry2 group mice. PcoA revealed an obvious separation between the microbiota community of FMT/shScramble group mice and FMT/shCry2 group mice, similar to the differences observed between the shScramble and shCry2 group mice (Fig. 4B).

Fig. 4.

Transfer of “shCry2 microbiota” induces cognitive decline, disruption of the intestinal barrier and BBB, increased tau pathology. (A) An overview of the experimental design. Three-month-old mice were subjected to a seven-week combined antibiotic treatment, followed by a three-day interval before conducting fecal microbiota transplantation experiments, which lasted for one week. Behavioral tests and molecular biology assays were performed after the transplantation., (B) Microbial communities clustered using PCoA of the Bray-Curtis Distance in shScr, shCry2, FMT/shScr, and FMT/shCry2 mice. The percentage of variation explained by the principal coordinates was indicated on the axes, (n = 6 mice/group)., (C-F) Escape latency of the FMT/shScr and FMT/shCry2 mice for 5 consecutive days in the MWM test, (C). Number of platform crossings in the spatial probe trail of the MWM test, (D). Percentage of time spent in the target quadrant in the spatial probe trail of the MWM test, (E). Representative heatmaps of the swimming path in the spatial probe trail of day 6 in the MWM test, (F), (n = 10–11 mice/group)., (G) Representative Western blot and quantification of Occludin and ZO-1 in the colon of the FMT/shScr and FMT/shCry2 mice., (H) H&E staining, (left), Alcian blue staining, (left), and representative ZO-1 immunofluorescence staining, (right) were performed on paraffin-embedded colon sections. Representative immunofluorescence staining, (right) against mucin-2 was performed on Carnoy-fixed colon sections. Magnification × 20. Scale bar = 100 μm., (I) Quantification of, (H)., (J-K) Representative Western blot and quantification of Occludin, ZO-1, Tau1, Tau5, pS396-tau and pT231-tau in the hippocampus of FMT/shScr and FMT/shCry2 mice. All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using a repeated measures 2-way ANOVA test with Bonferroni post hoc comparison in panel, (C), and the Student’s t-test in panel, (D, E, G, I, J, K). FI: Fluorescence intensity. shScr: shScramble

In the MWM test, the FMT/shCry2 group showed an increased escape latency during the training days, especially the second and fifth days (Fig. 4C). Moreover, compared to the FMT/shScramble group, the FMT/shCry2 group displayed a decrease in both the number of platform crossings and the time spent in the platform quadrant (Fig. 4D-F). Then, we determined the protein expression of Tau1, pT231-tau, pS396-tau, Tau5, ZO-1, and Occludin by Western blot. There was a significant upregulation in the protein expression of pT231-tau and pS396-tau, a downregulation in the expression of Tau1, and no significant difference in Tau5 expression in the hippocampus of FMT/shCry2 group mice as compared with the FMT/shScramble group (Fig. 4K). Additionally, the protein expression of ZO-1 and Occludin was decreased in both the hippocampus and colon in the FMT/shCry2 group mice compared to the FMT/shScramble group (Fig. 4G, J). Immunofluorescence analysis showed increased expression of pS396-tau while a decreased expression of ZO-1 in the hippocampus of the FMT/shCry2 group as compared with the FMT/shScramble group (Supplementary Fig. 3A-D). Alcian blue staining revealed a thinner and discontinuous mucus layer in the FMT/shCry2 group compared to the FMT/shScramble group. Immunofluorescence staining of ZO-1 and MUC2 illustrated the impairment of the intestinal mucus layer in the FMT/shCry2 group compared to the FMT/shScramble group (Fig. 4H, I). Thus, these results indicated that the disturbance of microbiota induced by hippocampal Cry2 deficiency could lead to cognitive impairment and tau-related pathology.

Effect of hippocampal Cry2 knockdown on the intestinal metabolites in mice

The gut microbiota (GM) produces various metabolites through metabolic activities, which play a crucial role in the host’s neural and immune activity. To further explore how Cry2 deficiency in hippocampus promoted metabolic activity changes, we conducted the UHPLC-Q-Exactive-Orbitrap-MS on the intestinal feces to quantify the metabolites (e.g., major metabolites in fatty acids, carboxylic acids and derivatives, indoles and derivatives) in the colon from shScramble and shCry2 group mice (Fig. 5A). Then we calculated the differentially enriched metabolites between the shScramble and shCry2 group mice. Among these metabolites, 224 metabolites were found to be upregulated significantly while 150 metabolites were downregulated (Fig. 5B) and the cluster heatmap displayed the 11 representative different metabolites expression between the shScramble and shCry2 group (Fig. 5C). By KEGG analysis, we identified that biosynthesis of unsaturated fatty acids, sphingolipid metabolism, sphingolipid signaling pathway, Glycosylphosphatidylinositol (GPI)-anchor biosynthesis, FoxO signaling pathway and so on involved in the metabolic process associated with Cry2 knockdown in hippocampus. Notably, sphingolipid metabolism and sphingolipid signaling pathways were the most significant pathways in this process (Fig. 5D).

Fig. 5.

Gut microbiota induces sphingolipid pathway disturbance in the peripheral and brain. (A) Representative metabolites classified and counted based on their chemical taxonomy in shScr and shCry2 mice. The size of the pie chart corresponded to the relative abundance levels of metabolites, (n = 10 mice/group)., (B) Volcano plot of differentially expressed metabolites from shScr versus shCry2 mice. Blue, red, and grey represented downregulated, upregulated, and no significantly expressed metabolites, respectively, (n = 10 mice/group)., (C) Unsupervised hierarchical clustering of differentially expressed metabolites from shScr and shCry2 mice, (n = 10 mice/group)., (D) Pathway enrichment analysis of differentially expressed metabolites between shScr and shCry2 mice, (n = 10 mice/group)., (E) Representative Western blot and quantification of Sphk1 and S1PR1 in the hippocampus of shScr and shCry2 mice., (F) LC/MS analysis of Cer(18:1/18:0) and Cer(18:2/18:0) in shScr and shCry2 mice., (G) Linear regression analyses between sphingosine and Akkermansia, (left plot) and Ruminococcaceae, (right plot) expression levels from data integrated from 16 S rRNA sequencing and metabolomic sequencing. All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using the Student’s t-test in panel, (E, F). shScr: shScramble

The crosstalk between the brain and the gut has long been appreciated in various neurodegenerative diseases. To investigate whether changes in the sphingolipid pathway in the peripheral could influence hippocampal sphingolipid metabolism, we examined the protein levels of Sphk1 and S1PR1 in the hippocampus using Western blot. Compared to the shScramble, shCry2 showed decreased protein expression of Sphk1 and increased expression of S1PR1 (Fig. 5E). Additionally, the FMT/shCry2 group mice exhibited a similar expression trend of Sphk1 and S1PR1 in the brain as compared with the FMT/shScramble group mice (Supplementary Fig. 4A). Furthermore, to quantify the expression features of various components of sphingolipids, UHPLC-timsTOF lipidomics profiles were generated on hippocampus samples by LC/MS. Results indicated that the expression of Cer(d18:2/18:0) and Cer(d18:1/18:0) decreased in the shCry2 group compared to the shScramble group, in line with the findings of metabolomics sequencing analysis (Fig. 5F).

We further assessed the correlation between the metabolites and the gut microbiota to explore the characteristics of the metabolites based on the Spearman correlation coefficient method in the shCry2 group compared to the shScramble group mice. Sphingolipid is positively correlated with AKKermansia, Helicobacter, Romboutsia, and Quinella but negatively correlated with Streptococcus and Ruminococcaceae. In our study, we found a positive correlation between sphingolipids and Akk, while a negative correlation was observed with Ruminococcaceae (Fig. 5G, Supplementary Fig. 4B). Taken together, these data suggested that hippocampal Cry2 deficiency significantly alters intestinal metabolite profiles, particularly affecting the sphingolipid metabolic pathway. The observed changes in peripheral sphingolipid metabolism are strongly correlated with hippocampal sphingolipid dynamics, highlighting a potential link between gut-derived metabolic processes and brain function.

FTY720 attenuated cognitive impairment and pathological changes induced by hippocampal Cry2 deficiency

Next, we used the sphingolipid-1-phosphate receptor 1 (S1PR1) agonist FTY720 targeting the Sphk1/S1PR axis to explore the molecular mechanism of Cry2 mediated pathological changes (Fig. 6A). As indicated in Fig. 6B, we observed that FTY720 treatment displayed shortened escape latency, increased time spent in the platform quadrant, and increased number of platform crossings in shCry2-FTY720 group as compared with the shCry2-saline group, and there was no significant difference between the shScramble-saline and shScramble-FTY720 (Fig. 6B-E). In addition, administration of FTY720 reversed the protein expression of Occludin, ZO-1, Sphk1, and S1PR1 in the hippocampus of the shCry2-FTY720 group as compared with the shCry2-saline group (Fig. 6F-G). FTY720 also significantly suppressed the protein expression of S1PR1 in shScramble-FTY720 compared to shScramble-saline (Fig. 6G). Compared to the shCry2-saline group, shCry2-FTY720 remarkably reversed the protein expression of pT231-tau, pS396-tau, and Tau1, while there was still no significant difference in Tau5 expression between the two groups (Fig. 6H). Immunofluorescence results showed that FTY720 treatment reversed the expression of pS396-tau in the hippocampus of shCry2-FTY720 group mice (Fig. 6I, J). Altogether, these findings indicated that FTY720 treatment could reverse cognitive decline, reduce tau-related pathological changes, alleviate cerebral Sphk1 expression, and repair the BBB in the hippocampus.

Fig. 6.

FTY720 ameliorated the cognitive decline, impairment of BBB and tau pathology. (A) An overview of the experimental design. After receiving stereotaxic brain injections, three-month-old mice were housed for six weeks, followed by intraperitoneal injections of saline or FTY720 for two weeks. Behavioral and molecular biology tests were then conducted., (B-E) Escape latency of the shScr-saline, shScr-FTY720, shCry2-saline, and shCry2-FTY720 mice for 5 consecutive days in the MWM test, (B). Percentage of time spent in the target quadrant in the spatial probe trail of the MWM test, (C). Number of platform crossings in the spatial probe trail of the MWM test, (D). Representative heatmaps of the swimming path in the spatial probe trail of day 6 in the MWM test, (E), (n = 11–12 mice/group)., (F-H) Representative Western blot and quantification of Occludin, ZO-1, (F), Sphk1, S1PR1, (G), Tau1, pS396-tau, Tau5, and pT231-tau, (H) in the hippocampus of shScr-saline, shScr-FTY720, shCry2-saline, and shCry2-FTY720 mice., (I, J) Representative immunofluorescence staining and quantification of pS396-tau, (red) in the hippocampus of shScr-saline, shScr-FTY720, shCry2-saline, and shCry2-FTY720 mice. Magnification × 10. Scale bar = 100 μm. All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using a repeated measures 2-way ANOVA test with Bonferroni post hoc comparison in panel, (B) and one-way ANOVA, (C-D, F-I). FI: Fluorescence intensity. shScr: shScramble

NLRP3 knockout reversed cognitive impairment and the pathological changes induced by hippocampal Cry2 deficiency

Then we tested the NLRP3 expression by Western blot, the protein expression of NLRP3 and IL-1β was elevated in the shCry2 group as compared with the shScramble (Fig. 7A). Similarly, increased protein expression of NLRP3 and IL-1β was observed in the FMT/shCry2 group compared to the FMT/shScramble group (Fig. 7B). Additionally, immunofluorescence results demonstrated activation of microglia in the shCry2 and FMT/shCry2 groups compared to the shScramble and FMT/shScramble groups, respectively (Fig. 7C-F). To explore the crucial role of NLRP3 in the pathological changes induced by hippocampal Cry2 deficiency, we performed Cry2 knockdown in the hippocampus of NLRP3 knockout (NLRP3^−/−) mice and their matched littermate WT mice. In the MWM test, NLRP3^−/−-shCry2 group mice showed shortened escape latency, increased time spent in the platform quadrant, and increased number of platform crossings as compared with WT-shCry2 group mice (Fig. 7G-J). Western blot analysis indicated that the expression of pT231-tau and pS396-tau decreased while the expression of Tau1 increased in the NLRP3^−/−-shCry2 group compared to the WT-shCry2 group (Fig. 7K). Immunofluorescence analysis results indicated that the expression levels of pS396-tau were significantly downregulated in the hippocampus of the NLRP3^−/−-shCry2 group compared to the WT-shCry2 group (Fig. 7L-M). What’s more, we examined the protein expression of NLRP3 and IL-1β in the shScramble-saline, shScramble-FTY720, shCry2-saline, and shCry2-FTY720. Administration of FTY720 reversed the expression of NLRP3 and IL-1β in the shCry2-FTY720 compared with shCry2-saline, and there was no significant difference between the shScramble-saline and shScramble-FTY720 (Fig. 7N). To explore how FTY720 alleviated the expression of NLRP3, we treated the BV2 cells with different concentrations of S1P receptor agonist FTY720, as well as S1PR1 antagonist W146. We induced an increased protein expression of NLRP3 in BV2 cells by adding LPS, and the expression of NLRP3 gradually decreased with the concentration of FTY720 increased, reaching the maximum reduction at a concentration of 10µM (Fig. 7O). In addition, the expression of NLRP3 was increased by administration of W146 at the concentrations of 1µM and 10µM by Western blot (Fig. 7P). Especially notable is the gradient change in NLRP3 protein expression at concentrations of 1µM and 10µM of W146, suggesting a potentially greater involvement of S1PR1 in this context. The above findings suggested that sphingolipid changes in the periphery induced an activation of the Sphk1/S1PR1/NLRP3 axis in the hippocampus through the impaired intestinal barrier and BBB.

Fig. 7.

NLRP3 knockout reversed cognitive impairment and the pathological changes induced by Cry2 deficiency. (A-B) Representative Western blot and quantification of NLRP3, IL-1β in the hippocampus of shScr, shCry2 (A), FMT/shScr, and FMT/shCry2 (B) mice. (C-D) Immunofluorescence staining of IBA-1 (red) in the hippocampus of shScr, shCry2 (C), FMT/shScr, and FMT/shCry2 (D) mice. Magnification × 10. Scale bar = 100 μm. Magnification × 20. Scale bar = 50 μm. (E-F) The relative area of soma and process length of microglia in shScr, shCry2, FMT/shScr, and FMT/shCry2 mice. (G-J) Escape latency of the WT-shCry2 and NLRP3−/−-shCry2 mice for 5 consecutive days in the MWM test (G). Number of platform crossings in the spatial probe trail of the MWM test (H). Percentage of time spent in the target quadrant in the spatial probe trail of the MWM test (I). Representative heatmaps of the swimming path in the spatial probe trail of day 6 in the MWM test (J) (n = 9–10 mice/group). (K) Representative Western blot and quantification of Tau1, Tau5, pS396-tau, and pT231-tau in the hippocampus of WT-shCry2 and NLRP3−/−-shCry2 mice. (L-M) Representative immunofluorescence staining and quantification of pS396-tau (red) in the hippocampus of WT-shCry2 and NLRP3−/−-shCry2 mice. (N) Representative Western blot and quantification of NLRP3, IL-1β in the hippocampus of shScr-saline, shScr-FTY720, shCry2-saline, and shCry2-FTY720 mice. (O-P) Representative Western blot and quantification of NLRP3 in BV2 cells treated with LPS, FTY720, and W146. All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using a repeated measures 2-way ANOVA test with Bonferroni post hoc comparison in panel (G), the Student’s t-test in panel (A-B, E-F, H-I, K, M), and one-way ANOVA (N-P). FI: Fluorescence intensity. shScr: shScramble

Hippocampal Cry2 overexpression alleviated tau pathology and improved memory deficits in the AD mouse model

To investigate when the Cry2 expression began to decrease in the hippocampus of the APP/PS1 mouse model of AD, we examined the protein expression in mice aged 2, 3, 4, 5, and 6 months. Western blot result indicated that the protein expression of CRY2 began to decrease from 6 months, so we overexpressed the Cry2 in the hippocampus of 4.5-month-old APP/PS1 mice for 6 weeks and then performed the MWM test (Fig. 8A). Expression of CRY2 was confirmed by immunofluorescence assay in the hippocampus of APP/PS1 mice (Supplementary Fig. 5B). In the MWM test, APP/PS1-Cry2^OE group mice displayed shortened escape latency, increased time spent in the platform quadrant, and increased number of platform crossings as compared with APP/PS1-Ctrl group mice (Fig. 8B-E). Compared to APP/PS1-Ctrl, APP/PS1-Cry2^OE showed a reduced protein expression of pT231-tau, pS396-tau, pT181-tau, pT217-tau, and an increased expression of Tau1. Additionally, there was no significant difference in Tau5 between the two groups (Fig. 8F-G). Immunofluorescence analysis revealed that the pS396-tau expression levels were reduced in the APP/PS1-Cry2^OE group compared to the APP/PS1-Ctrl group (Fig. 8H). These results indicated that Cry2 overexpression in the hippocampus of APP/PS1 prevented memory loss and alleviated the tau-related pathology.

Fig. 8.

Overexpressing Cry2 alleviates the memory deficits and tau pathology. (A) Representative Western blot and quantification of CRY2 in the hippocampus of APP/PS1 mice at 2, 3, 4, 5, and 6 months of age. (B-E) Escape latency of the APP/PS1-Ctrl and APP/PS1-Cry2OE mice for 5 consecutive days in the MWM test (B). Percentage of time spent in the target quadrant in the spatial probe trail of the MWM test (C). Number of platform crossings in the spatial probe trail of the MWM test (D). Representative heatmaps of the swimming path in the spatial probe trail of day 6 in the MWM test (E) (n = 7-8 mice/group). (F-G) Representative Western blot and quantification of Tau1, Tau5, pS396-tau, pT231-tau, pT181-tau, and pT217-tau in the hippocampus of APP/PS1-Ctrl and APP/PS1-Cry2OE mice. (H) Representative immunofluorescence staining and quantification of pS396-tau (red) in the hippocampus of APP/PS1-Ctrl and APP/PS1-Cry2OE mice. All data were expressed as mean ± SEM. Data were considered significant if *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Means were compared using a repeated measures 2-way ANOVA test with Bonferroni post hoc comparison in panel (B), one-way ANOVA (A), and the Student’s t-test in panel (C, D, G, H). FI: Fluorescence intensity. shScr: shScramble

Discussion

The circadian clock is an internal rhythm system in the body that regulates the sleep-wake cycle and other physiological processes. AD patients typically exhibit sleep disturbances and daytime sleepiness, which are associated with circadian clock dysfunction [44]. The precise mechanism between the circadian clock gene Cry2 and cognitive function remains poorly understood, with limited studies addressing this relationship to date. In the present study, we found that reduced hippocampal Cry2 expression led to alterations in the composition of gut microbiota, sphingolipid metabolism disturbance, compromised intestinal barrier and BBB integrity, cognitive decline, and the development of tau pathology. Our findings demonstrated that Cry2 deficiency in the hippocampus leads to cognitive impairment through the gut-brain axis mediated S1P/NLRP3/IL-1β pathway.

Previous study indicated that mice lacking both Cry1 and Cry2 show impaired recognition memory, while depression-related behaviors are largely unaffected [45]. In addition, Cry1/2 deficient mice exhibited mild cognitive and social deficits and heightened anxiety, accompanied by increased amygdala cFos activation following anxiogenic stimuli [46]. Consistent with these observations, our data specifically demonstrated that hippocampal Cry2 deficiency led to cognitive decline, as evidenced by increased escape latency and reduced time spent in the target quadrant during the MWM test. Moreover, Cry2 had been previously reported to modulate sleep deprivation-induced cognitive decline in the AD mouse model through the CISH–STAT1 signaling axis and maintenance of synaptic integrity [47]. In agreement with our findings, the authors of that study also suggested that Cry2 may play a crucial role in preventing sleep deprivation-induced cognitive decline in the AD mouse model.

Thasis. et al. demonstrated that the microbiota displayed diurnal oscillations influenced by feeding rhythms in both mice and humans, resulting in time-specific compositional and functional profiles over the course of a day. Moreover, the gut microbiota loses its rhythmicity in Per1/Per2 deficient mice [17]. Additionally, our previous study revealed that chronic sleep deprivation induced cognitive decline via dysbiosis of the intestinal microbiota [48], prompting our focus on the gut microbiota dysbiosis induced by Cry2 deficiency in the hippocampus. In our present study, we observed that the abundance of gut microbiota was higher in shCry2 mice compared to controls. Firmicutes, Bacteroidetes, Desulfobacterota, and Proteobacteria displayed alterations in shCry2 mice, which have also been frequently reported in studies of AD [49, 50]. Additionally, at the level of family, the species diversity of Lachnospiraceae and Ruminococcaceae significantly increased in shCry2 mice as compared with control mice, consistent with the previous study indicating that the increasing abundance of Lachnospiraceae and Ruminococcaceae was associated with aging and cognition [51]. To investigate the deleterious role of dysbiosis of intestinal microbiota in hippocampal Cry2 deficiency mice, we transplanted the fecal microbiota from the shScramble and shCry2 groups into antibiotic-treated mice, respectively. The FMT/shCry2 group mice showed a cognitive decline, increased tau-related pathology, impaired intestinal barrier and BBB, also indicated by decreased expression of ZO-1 and Occludin, and a discontinuous mucus layer. The pathological changes observed in FMT/shCry2 mice were reminiscent of those in shCry2, underscoring the crucial role played by alterations in intestinal microbiota in this process.

An increasing number of studies have confirmed that microbial-derived metabolites can act as molecular or chemical messengers mediating interactions between hosts and microorganisms [52, 53]. Metabolomics analysis demonstrated relatively low levels of sphingosine in the shCry2 group as compared with the shScramble group, and it was indicated that the sphingolipid metabolism pathway was participated in it. Sphingolipids are essential lipids formed by the N-acylation of a sphingolipid backbone with various fatty acids, giving rise to a variety of ceramide species. What’s more, sphingosine, ceramide (Cer), and their respective 1-phosphates (C1P and S1P) are the four main bioactive sphingolipid forms in the brain, serving as intracellular second messengers [54, 55]. Cer served as one of the precursors for the generation of sphingosine by ceramidases, then sphingosine kinases (Sphk1, Sphk2) phosphorylate sphingosine into S1P in various cellular compartments, and S1P subsequently exerts its effects by binding to S1P receptors (S1PR1-5) [56]. In our study, we found that the protein expression of Sphk1 decreased and S1PR1 expression increased in the hippocampus of shCry2 as compared with shScramble. Additionally, the Cer (18:1/18:0) and Cer (18:2/18:0) were reduced in the shCry2 group compared to the shScramble group, indicating a disturbance of the sphingolipid pathway in the brain. Integrated Metabolomics and 16 S rRNA analysis indicated sphingosine is positively related to Akkermansia and negatively related to Ruminococcaceae. Compared to the shScramble group, the shCry2 group showed increased expression of Ruminococcaceae and decreased expression of Akkermansia, which is consistent with the decline in sphingosine levels in the colon of shCry2 mice. Akkermansia muciniphila (Akk), a genus of Verrucomicrobia, has been shown to restore the damaged integrity of the intestinal epithelium barrier, improve insulin resistance, and reduce endotoxin levels. It could alleviate the characteristic pathology of AD and repair the impairment of memory and spatial learning in the AD mouse model [57], which is in line with our results. In addition, decreased expression of Ruminococcaceae could serve as predictive markers for the rapidly progressive MCI [58]. These data suggested that the Sphk1/S1PR1 signaling pathway is a core pathway in cognitive decline mediated by intestinal microbiota dysbiosis in shCry2 mice.

The S1P signaling pathway has important and diverse functions, and S1P receptors are involved in regulating pathological processes such as lymphocyte trafficking, brain and cardiac functions, vascular permeability, and bronchial tone [59]. FTY720, an S1P receptor modulator (except for S1PR2), is a sphingosine structural mimetic, which could penetrate the BBB. After phosphorylation in the tissue, FTY720 becomes an analog of S1P, then binds to S1PR1 and exerts receptor internalization effects [60]. In 5×FAD mice, FTY720 could reverse the expression of Sphk1 and S1PR1 in the hippocampus [61]. We found that FTY720 could reverse the pathological and destructive memory and cognitive ability, accompanied by an increased level of Sphk1 and a decreased level of S1PR1 in shCry2 mice, which highlights the S1P/S1PR1 pathway in the brain induced by intestinal microbiota dysbiosis plays a determinative role in shCry2 mice. Furthermore, Sphk1 was shown to be positively associated with the gene expression of NLRP3 in protracted bacterial bronchitis and bronchiectasis [62]. S1P could induce the activation of the NLRP3 inflammasome in LPS-primed BMDMs by activating Caspase-1, promoting IL-1β maturation, facilitating the formation of apoptosis-associated speck-like protein containing a CARD (ASC) speck, and inducing IL-1β secretion [63]. Thus it is hypothesized that there exists a close relationship between S1P and NLRP3.

NLRP3 is primarily expressed in microglial cells and triggers an inflammatory response, leading to the activation of the downstream inflammatory cytokine IL-1β. Mounting evidence indicates that the neuroinflammation caused by the activation of NLRP3 inflammasome aggravates the tau pathology and induces neuronal damage and cellular apoptosis in the AD mouse model [64, 65]. We also investigated whether dysbiosis of intestinal microbiota could lead to increased neuroinflammation caused by Cry2 deficiency in the hippocampus. Our data revealed that evaluated expression levels of NLRP3 and IL-1β, along with heightened microglial activation in the hippocampus of both shCry2 mice and FMT/shCry2 mice. Knockdown of Cry2 in NLRP3^−/− mice displayed improved cognitive memory loss and cognitive ability. Additionally, FTY720 treatment reversed the expression of NLRP3 and IL-1β, suggesting a potential link between S1P pathway activation and NLRP3/IL-1β pathway activation. The S1PR1 agonist FTY720 reduced the expression of NLRP3 and IL-1β [66]. Our findings also confirmed that S1PR agonist FTY720 and S1PR1 antagonist dose-dependently modulated the expression of NLRP3 inflammasome, thereby confirming the upstream-downstream relationship between S1PR and NLRP3.

Studies have shown that two-month-old AD mice already display altered activity patterns compared with WT mice, accompanied by significant changes in the expression of core circadian clock genes in the SCN [67]. In our study, we observed that CRY2 protein expression began to decline at 6 months of age. Furthermore, mounting evidence underscores the critical role of dysregulated clock gene expression as a driver of both cognitive decline and AD-like pathology. Amyloid-β is generated by the sequential proteolytic cleavage of the amyloid precursor protein (APP) by β-secretase and γ-secretase [68]. Aβ accumulation represents a central pathological mechanism in AD, and studies demonstrate that the circadian rhythm system actively modulates its production and clearance. Mechanistically, REV-ERBα/β act as negative regulators of microglial Aβ clearance, and their inhibition enhances BMAL1-driven phagocytic activity and mitigates amyloid pathology in 5×FAD mice [69]. In the 5×FAD model, LD (light-dark) interacted with the central circadian clock gene to exacerbate Aβ pathology, whereas deletion of Bmal1 in GABAergic neurons weakens these rhythms and consequently reduces APP processing, Aβ plaque burden, peri-plaque p-tau, and induces extracellular matrix (ECM) gene expression under LD conditions [70]. Our results also demonstrated that hippocampal Cry2 overexpression in APP/PS1 mice alleviated the memory deficits and attenuated tau pathology. The statistical difference in spatial learning performance was observed on the third day, and an improving trend was observed on day 4 and days 5 during the training phase in MWM. Previous studies suggest that 6-month-old APP/PS1 mice may not yet display pronounced cognitive deficits in the MWM compared to control groups, which may explain why the mice did not show a marked difference in MWM learning performance at this stage [71, 72]. Consequently, the observed pattern is interpreted as an enhancement of early learning and memory consolidation rather than a rescue from established, severe cognitive impairment. Future studies should therefore consider increasing the sample size and prolonging the Cry2 intervention period to fully elucidate the therapeutic trajectory.

In summary, altered gut microbiota and sphingolipid metabolites are the initiation factor for hippocampal Cry2 deficiency, which induces intestinal barrier disruption. The peripheral sphingolipid disturbance through the impaired BBB could further influence the synthesis of sphingolipid and activate NLRP3 inflammasome in the CNS, which aggravates tau pathology and cognitive impairment (Fig. 9). Our findings identify a highly detrimental role of S1P/S1PR1/NLRP3 signaling in the pathological changes induced by hippocampal Cry2 deficiency. Cry2 and its involved passway might serve as a potential therapeutic target for the treatment of AD.

Fig. 9.

Hippocampal Cry2 deficiency leads to cognitive impairment through the gut-brain axis mediated S1P/NLRP3/IL-1β pathway. Hippocampal Cry2 deficiency induces gut dysbiosis, intestinal barrier disruption, and altered intestinal sphingolipid metabolism, which in turn leads to dysregulated sphingolipid metabolism in the hippocampus, elevated production of NLRP3 inflammasome, and activation of the S1P/NLRP3/IL-1β signaling pathway, ultimately resulting in cognitive impairment and exacerbated tau pathology

Conclusion

Our study reveals a critical role of the circadian clock gene Cry2 in the pathogenesis of AD. We demonstrated that hippocampal Cry2 deficiency contributes to cognitive impairment by disrupting gut microbiota and activating the S1P/NLRP3/IL-1β pathway within the brain. These findings underscore the significance of the gut-brain axis in AD and suggest that targeting the Cry2-related pathways, specifically through the modulation of S1P signaling, may offer new therapeutic opportunities. Further research is needed to deepen our understanding of Cry2’s role in the gut-brain axis and to explore its potential as a target for therapeutic intervention in AD.

Abbreviations

AD

Alzheimer’s disease

SCN

Suprachiasmatic nucleus

TTFL

Transcription-translation feedback loop

Clock

Circadian locomotor output cycles kaput

Bmal1

Brain and muscle Arnt-like protein 1

Per

Period

Cry

Cryptochrome

BBB

Blood-brain barrier

SLs

Sphingolipids

Sphk1

Sphingosine kinase1

SPMs

Specialized proresolving mediators

S1P

Sphingosine-1-phosphate

DAMP

Damage-associated molecular pattern

Sph

Sphingosine

LefSe

Linear discriminant analysis Effect Size

LDA

Linear Discriminant Analysis

MUC2

Mucin-2

GM

Gut microbiota

ASC

Apoptosis-associated speck-like protein containing a CARD

Akk

Akkermansia muciniphila

Sphk

Sphingosine kinase

Cer

Ceramide

S1PR1

Sphingolipid-1-phosphate receptor 1

MWM

Morris water maze

GFP

Green fluorescent protein

AAV

Adeno associated virus

NLRP3

NLR Family Pyrin Domain Containing 3

Authors’ contributions

F.G., N.Z., and Q.G.R. designed and conceptualized the project. F.G. and N.Z. performed the experiments and wrote the manuscript. L.Z. and N.Z. analyzed the data. X.T.L., X.C., and Z.T.W. helped interpret the results. Z.J.Z. and Q.G.R. contributed to the experiment’s performance and experiment design. All authors read and approved the final manuscript.

Funding

The work was supported by the National Natural Science Foundation of China under grant [81870850], the Postgraduate Research & Practice Innovation Program of Jiangsu Province under grant [KYCX23_0322], and Zhongda Hospital Affiliated to Southeast University, Jiangsu Province High-Level Hospital Construction Funds under grant [GSP-LCYJFH07].

Data availability

The GSE138260, GSE5281, GSE29378, and GSE48350 datasets are available on the GEO datasets from NCBI. 16 S rRNA sequence raw data have been deposited in the NCBI database under accession number PRJNA1120391. Metabolomics and lipidomics raw data have been deposited in the China National Center for Bioinformation database under accession numbers [PRJCA026908](https:/ngdc.cncb.ac.cn/gsub/submit/bioproject/PRJCA026908) (OMIX010084) and PRJCA039884 (OMIX010092). The code used to analyze the clock genes is available on GitHub [https://github.com/19960728/Cry2-paper].

Declarations

Ethics approval and consent to participate

All animal-related procedures were performed in accordance with ethical standards and were approved by the Institutional Animal Care and Use Committee of Southeast University (20210924075).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.