Thirteen 5-week-old Sprague-Dawley (SD) male rats (130–150 g) purchased from the Dashuo company were housed in plastic cages in a temperature-controlled (25°C) colony room at a 12/12 h light/dark cycle. Food and water were available ad libitum. All animal procedures in this study were approved by the Ethics Committee of West China Hospital of Sichuan University (WCHSIRB-D-2021-009) and conformed to the ARRIVE (Animal Research: Reporting of in vivo Experiments) guidelines for preclinical studies. All efforts were made to minimize the number of animals used.

After 1 week of environment acclimation, the rats were randomly divided into three groups, AD + ligation + non (AD, n = 4), AD + ligation + F. nucleatum (AD + F.n, n = 5), and blank (Con, n = 4). The rats were subcutaneously injected in the neck and back with D-galactose and AlCl₃ (120 mg/kg/day and 10 mg/kg/day, 0.1–0.5 ml) or PBS every day from the 6th week to the 22nd week to establish an AD-like model (as shown in Figure 1A). Rats were ligated with a 5–0 silk suture around the bilateral maxillary second molars (M2) for experimental periodontitis in the 16th week. F. nucleatum ATCC 25586 was anaerobically grown and resuspended to a concentration of 1 × 10⁹ CFU/mL in 4% carboxymethyl cellulose (CMC), and was smeared on the silk suture (0.2 ml) every other day from the 16th week to the 22nd week (Chukkapalli et al., 2014). The Con group received 0.2 ml of PBS in 4% CMC.

The Morris water maze test (MWMT, MT-200, Chengdu Taimeng Technology Co., Ltd.) was performed in the 23rd week to study the spatial learning and memory abilities of rats. Morris water maze test (MWMT) was carried out as described by Bhuvanendran et al. (2019) with few modifications. The water pool was a circular white pool with a diameter of 200 cm and a depth of 60 cm filled with non-toxic white paint opaque water and the temperature was held constant at 23 ± 1°C. The pool was separated into four equal quadrants, and an invisible platform was submerged 1 cm under the water surface in the center of the first quadrant and was considered the target quadrant. During the training days, the location of the platform remained the same and the rats were given tests for successive 4 days. In each trial, a rat was released in a quadrant and allowed to swim freely for 90 s to find the invisible platform and stay on it for 15 s. If the rats could not find the located platform in 90 s, it was gently guided to the platform, and allowed to stay there for 15 s. The time taken to reach the platform (escape latency) was recorded and the practice was repeated for four different starting quadrants of the trial per day. A probe trial test was performed on the 5th day without the platform to evaluate the memory consolidation rate. The rats were allowed to swim freely for 90 s, and count the number of times the mouse crossed the area where the platform had been placed.

Animals were sacrificed one day after the probe trial test. First, fresh feces of rats were collected for quick freezing. The rats were deeply anesthetized with an intraperitoneal injection of 2.5% tribromoethanol. After the collection of blood, the rats were sacrificed and the brain was collected. One cerebral hemisphere was immediately stored at −80°C until further analysis and the other half fixed in 4% paraformaldehyde for 24 h, then were processed and embedded in paraffin. Bilateral maxillary bones were removed and fixed in 4% paraformaldehyde for 24 h.

To evaluate morphological changes in the alveolar bone, the maxillary bone was scanned by Micro-CT (Jia et al., 2019). Fixed maxillary bones of each group were randomly selected and scanned with a Micro-CT (μCT50; SCANCO) at the voxel resolution of 10 μm. Three-dimensional reconstruction and data analysis were performed by Scanco Evaluation. To calculate bone loss, the distance between the cementoenamel junction and alveolar bone crest (CEJ-ABC distance) was measured for two predetermined maxillary sites on palatal sides of M2.

The histological changes in the sections of brain tissues from the different groups were observed using HE staining (Chen S. et al., 2017). After being preserved in 10% formalin solution for 24 h, brains were embedded in paraffin and 5 μm sections were sliced. The samples were then dewaxed and rehydrated, subsequently stained with hematoxylin and eosin, and examined under standard light microscopy for a general histopathology examination. The images were captured by a camera (Nikon, 90i, Tokyo, Japan).

The immunohistochemical staining was performed as described by Shin et al. (2019) with slight modifications (Shin et al., 2019). After being fixed in 4% paraformaldehyde for 24 h, the brains were embedded in paraffin and then cut at a thickness of 5 μm. The sections for staining were deparaffinized and washed. Following, they were heated in 0.01 M sodium citrate buffer (pH 6.0) for antigen retrieval. Endogenous peroxide activity was quenched with 3% hydrogen peroxide. Sections were incubated with primary antibodies against Aβ1–42 (Abcam, ab10148, 1:200) overnight at 4°C, followed by anti-rabbit IgG (ZSGB-BIO, PV-9001). Then sections were incubated with horseradish enzyme labeled streptomycin working solution at room temperature for 30 min. Finally, color was developed with 3,3′-diaminobenzidine (ZSGB-BIO, ZLI-9018), and then it was counterstained with hematoxylin. After gradient dehydration with alcohol, the slices were sealed with neutral resin. The images were captured. And 6 visual fields per slices were being counted in a blind manner.

The supernatant of brain protein was mixed with sample buffer and boiled. Protein from boiled samples was separated by 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and transferred to PVDF membranes. Then, the membranes were blocked in PBS–0.1% Tween 20 containing 5% milk and incubated with rabbit monoclonal IgG antibodies against GAPDH (1:1,000, Signalway Antibody, United States), Tau (phosphoThr181, 1:500, Signalway Antibody, United States), and mouse monoclonal IgG antibodies Tau (Tau46, 1:500, CST, United States) overnight at 4°C. The membranes were then incubated with goat anti-rabbit or rabbit anti-mouse IgG antibodies (1:5,000, Signalway Antibody, United States). The protein was visualized with Enhanced ECL Reagent Kit (Beijing Bio Excellence Biotechnology Co., Ltd., China). Images were captured by using a gel imaging system (Bio-Rad, United States).

Serum was isolated by centrifuging the blood after clotting at 4,000 rpm for 10 min. The serum level of LPS was measured with Tachypleus Amebocyte Lysate (EC64405, Xiamen Limulus Reagent Biotechnology Co., Ltd., xiamen, China) according to the manufacturer’s protocol.

To confirm the spread of periodontal pathogens from the mouth to the brain of AD-like rats, genomic DNA was isolated from the brains of all groups following the manufacturer’s protocol (DNeasy Blood and Tissue Kit, Qiagen, Germany) (Wang et al., 2015). DNA amplification was performed using a PCR amplification kit (Sangon Biotech Co., Ltd., Shanghai, China) according to the manufacturer’s instructions. Briefly, the PCR mixture contained 12.5 μl Taq PCR Master, 0.5 μl (10 μg/ml) DNA samples, 1 μl forward primer, 1 μl reverse primer, and 1 μl sterilized ddH2O. The primers used for amplification were as follows: 5′-GGCCACAAGGGGACTGAGACA-3′ (forward) and 5′-TTTAGCCGTCACTTCTTCTGTTGG-3′ (reverse) (Sangon Biotech Co., Ltd., Shanghai, China) and its reliability is verified. The reaction temperature was 94°C for 4 min, followed by 40 cycles of 94°C for 30 s, 60°C for 30 s, 72°C for 10 s, and then 72°C for 10 min. DNA amplification products were electrophoresed in a 2.0% agarose gel electrophoresis under 120 V for 20 min with F. nucleatum ATCC 25586 DNA as the positive group (183 bp) and a blank reaction system as the negative group.

DNA from 13 fecal samples of Con, AD, and AD + F.n groups were extracted using a Qiagen stool DNA extraction kit (Qiagen, Germany) according to the instructions. Eleven DNA samples (3 in Con, 3 in AD, 5 in AD + F.n), with ratios of 1.8–2.0 (for A260/280 nm) and > 1.8 (for A260/A230 nm), were used for downstream experiments (Wang et al., 2016). For the analysis of the composition of the gut microbiota, the V3 and V4 regions of the 16S rRNA gene were amplified using universal primers 338F and 806R. The sequencing was performed on the Illumina Miseq platform following the usual operating procedures of Personal Biotechnology Co., Ltd. (Shanghai, China). The DNA fragments were paired-end sequenced with the Illumina platform. The obtained sequences were denoise, quality controlled, dereplication, etc. to obtain high-quality sequences. The cluster_size module was used to cluster high quality sequences at a 97% similarity level, and the Operational taxonomic unit amplicon sequence variants (ASVs) tables were obtained. Finally, the singletons ASVs (ASVs detected in only 1 sample) and their representative sequences were removed from the ASV table. Calculated the length distribution of the high-quality sequences contained in the sample, and the obtained sequences were annotated with the Greengenes database. α diversity indices were determined using the Simpson index for diversity and the Chao1 index for species richness. β diversity can be demonstrated by Principal coordinate analysis (PCoA) and the Bray-Curtis distance was used to denote the β diversity distance (Li et al., 2019).

All data were statistically analyzed by GraphPad Prism 6.0. Data were expressed as mean ± SEM. Morris’s water maze was analyzed with repeated measurements ANOVA. Comparisons between the three groups were evaluated by ANOVA or Kruskal-Wallis Test. Differences with P < 0.05 were considered statistically significant.

Micro-CT showed that the AD and AD + F.n groups had significantly more alveolar bone loss than the Con group, indicating the successful induction of periodontitis. Furthermore, alveolar bone loss in the AD + F.n group in rats infected with F. nucleatum for 6 weeks was increased compared with those of rats in the AD group (Figures 1B,C). These results suggested that oral infection with F. nucleatum induced more alveolar bone loss in AD-like periodontitis rats.

To clarify whether F. nucleatum could induce cognitive impairment, we tested the learning ability of rats by MWMT. As shown in Figure 2A, the escape latency of the AD and AD + F.n groups were significantly increased from day 2 to day 4 when compared with the Con group (P < 0.05), indicating successful induction of the AD-like rat model through D-galactose and AlCl₃ injection. In addition, when compared with the AD group, the escape latency of the AD + F.n group was significantly increased, in particular, there was an average increase of about 20 s on the fourth day (P < 0.05). These data indicated that chronic F. nucleatum oral infection could damage the learning ability further of AD-like with periodontitis rats.

In the space exploration experiment, we found that the track of rats in the Con group was mainly concentrated in and around the target quadrant (the first quadrant), while rats in the AD and AD + F.n groups moved disorderly or presented marginal movement, and rarely enter the target quadrant (Figure 2C). Moreover, the AD and AD + F.n groups had a significantly decreased number of times cross the platform than the Con group. These data also indicate the successful construction of the AD-like rat model. However, we didn’t find any evidence that F. nucleatum aggravated memory ability in rats as the AD + F.n group had a similar number of times cross the platform with the AD group (Figure 2B).

To clarify whether F. nucleatum could induce neurodegeneration, the neuronal morphology in the cortex (Cotx) and hippocampus (Hipp) of rats by HE were first investigated. We found that neurons in Cotx and Hipp of rats in the Con group were neatly arranged and compact, without significant cell vacuolation and necrosis (Figure 3A). Neuronal degeneration and karyopyknosis in brain tissue were observed in AD and AD + F.n groups, and cells in the hippocampus were loosely organized and vacuolation.

The Aβ1–42 expression in Cotx and Hipp of rats was examined by IHC. And the number of positive Aβ1–42 cells in brain tissue of AD and AD + F.n groups significantly increased (Figures 3B,C, P < 0.05). Compared with the AD group, the expression of Aβ1–42 in the AD + F.n group was significantly increased by about one-fourth (P < 0.05). These data indicated that chronic F. nucleatum oral infection could trigger neurodegeneration and Aβ1–42 express in the Cotx and Hipp of AD-like rats.

The expression of p-Tau181 and Tau protein in the brain of rats were detected by WB, as shown in Figures 3D,E. Compared with the Con group, the level of p-Tau181 in the AD group and AD + F.n group was significantly increased (P < 0.05). In addition, when compared with the AD group, the relative expression of p-Tau181 in AD + F.n group was significantly increased by about one-third (P < 0.05). These data indicated oral infection with F. nucleatum could induce phosphorylated tau increased in the brain of AD-like periodontitis rats.

We next explored the possible mechanisms by which F. nucleatum-caused cognitive impairment, neurodegeneration, and Aβ1–42 expression. PCR was used to examine whether F. nucleatum entered the brain first. None of the brain tissue lysates demonstrated DNA from F. nucleatum in the non-infected and infected groups (data not shown).

Serum LPS of rats in each group was detected by LAL assay (as shown in Figure 4). Compared with the Con group, serum LPS of the AD group was significantly increased, suggesting the AD-like model could induce an increase in serum LPS levels, which may be related to systemic inflammation. Moreover, the AD + F.n group also had a significantly increased LPS level when compared with the AD group. These data suggested that F. nucleatum oral infection could trigger chronic systemic inflammation, we inferred that F. nucleatum might trigger neuropathy by inducing the release of LPS and the resulting inflammatory response in the circulatory system.

A total of 866,530 high-quality sequences were captured from 11 fecal samples, with an average of 78,775 sequences per sample, and average length of 426 bp. The gut microflora of all samples was classified into 29 phyla, 89 classes, 158 orders, 280 families, 470 genera, and 5,763 ASVs.

α-diversity index of gut microflora was evaluated according to the observed species, Simpson, and Chao1 indices. As shown in Figure 5A, there was no significant difference in the three parameters among the three groups (P > 0.05). Regarding gut microflora structure, the dissimilarity tests showed that it was different between the AD group and the Con group to some extent (P < 0.1), but the AD group was similar to the AD + F.n group (P > 0.05; as shown in Table 1). The rarefaction analysis revealed that the sequence depth was almost sufficient to recover the diversity of this community (Figure 5B). Venn figure showed that there were, respectively, 1,573, 2,789, and 3,082 ASVs in Con, AD, and AD + F.n groups. Among them, AD and AD + F.n overlapped the most ASVs, 407. Only 187 ASVs were shared by three groups (Figure 5C).

At the level of phylum (Figure 6A), the top five phylum were Firmicutes, Proteobacteria, Actinobacteria, Bacteroidetes, and TM7. The relative abundance in the top five phylum had no significant differences between groups. According to hierarchical clustering analysis, the samples from the Con group (A1–3) clustered together and were well separated from the AD group (B1–3). Moreover, the AD + F.n group (C1–5) had a relatively nearer distance to AD (B1–3) compared to Con, which meant their community was more similar to each other. Although some genera were common to all samples, such as Aerococcus, the variability in genera distribution among different samples was noticeable (Figure 6B). Then, the structure and predominant taxa from the phylum to genus level of the gut microbial community in each group was evaluated by the linear discriminant analysis effect size (LEfSe) test and displayed in Figure 6C. Marked taxa from AD was g_Coprococcus, and most specific taxa were from AD + F.n group in phylum and genus were p_Tenericutes and g_Prevotella.

At the genus level, Psychrobacter, Enterococcus, and Aerococcus accounted for 50% of the community in the Con group, while in the AD group Aerococcus constituted 81.98%, AD + F.n group consists of the genera of Aerococcus and Streptococcus for 50% of the community. For the top 20 genera (Figure 6D), a variance analysis was performed and found that Psychrobacter, Streptococcus, Lactococcus, Staphylococcus, Jeotgalicoccus, and Prevotella have a differential expression. When compared with the AD group, the relative abundance of Streptococcus and Prevotella in the AD + F.n group was significantly increased.

PICRUSt2 was used to predict metabolic pathway in MetaCyc and found that some pathways were significantly increased or decreased in the AD + F.n group compared to the AD group (Figure 6E). Furthermore, pathways related to neurotransmitter degradation were significantly changed in the gut microbiome of F. nucleatum-treated rats compared with AD. ORNARGDEG-PWY, ARGDEG-PWY, THREOCAT-PWY, PWY-7456, and PWY-6572 involved in the degradation of L-arginine, L-ornithine, putrescine, 4-aminobutanoate, L-threonine, mannan, and chondroitin sulfate were significantly increased, while CRNFORCAT-PWY involved in the creatinine degradation decreased.