Porphyromonas gingivalis may promote type 2 diabetes with memory problems by changing gut bacteria and metabolism

Forty SPF Male 4-week-old C57BL/6J mice were purchased from Chengdu Dossy Animal Experimental Co., LTD (Chengdu, China). All experiments were approved by the Ethics Committee of Chongqing General Hospital. The mice were randomly divided into normal and T2DM model groups. In the first four weeks, mice in the normal group were fed an ordinary diet, and T2DM model mice were fed a high-fat diet (D12492, Biopike, China) for 4 weeks. After fasting overnight, mice in the model group were intraperitoneally injected with 50 mg/kg streptozotocin (STZ, S8050, Solarbio, China), while mice in the normal group were intraperitoneally injected with the same amount of citric acid buffer. After 1 week, fasting plasma glucose (FBG) in the mice was detected, and FBG ≥ 11.1 mmol/L was confirmed as successful T2DM modeling. The model group was further divided into T2DM group and Pg + T2DM group, and the normal group was divided into control group and Pg control group (Pg), with 6 mice in each group. The Pg group and the Pg + T2DM group were injected with 200 µL 1.0 × 10⁸ CFU Pg solution (ATCC33277, Jinbio, China) in the tail vein of normal and T2DM mice, respectively(Sasaki et al. 2018). The administration was given once on Monday, Wednesday, and Friday each week, 3 times a week for 4 weeks, and the other groups were injected with 200 µL PBS. Random blood glucose was monitored weekly, oral glucose tolerance tests (OGTT) and insulin tolerance tests (ITT) were conducted, and the Morris water maze test and novel object recognition were performed to evaluate cognitive function after the experiment. After the behavioral experiment, the mice were sacrificed the next day for cervical dislocation, and brain tissue, blood, and colon contents were collected for follow-up experiments. The experimental arrangement was shown in Fig. 1.

The brain tissue sections were treated with 3-Aminopropyltriethoxysilane (APES) to prevent them from falling off and being dewaxed, then stained with 1% toluidine blue aqueous solution (G1032, Servicebio, China) preheated to 50℃ at 56℃ for 20 min, washed with distilled water and soaked in 70% alcohol for 1 min. After 95% alcohol was used for differentiation, gradient alcohol was dehydrated, xylene transparent, and neutral gum was sealed. Finally, the slices were observed and photographed under a microscope (BA210Digital, Panthera, China). The Nissl bodies in the hippocampal CA1 region were counted in three 400× field images collected from each sample. The Nissl bodies were dark blue and the background was colorless.

The SYN, Iba1, and Olfr78 in the CA1 region of the mouse hippocampus were detected by immunofluorescence. After paraffin sections were dewaxed to water, the antigen was repaired by microwave oven with citrate buffer (pH 6.0) for 20 min. The endogenous peroxidase was blocked when incubated with 3% hydrogen peroxide at room temperature for 25 min without Light. After bovine serum albumin was blocked for 30 min, the primary antibodies SYN (1:100; IPB6304, Baijia, China), Iba1 (1:100; GB12105-100, Servicebio, China) and Olfr78 (1:100; PA5-99482, ThermoFisher, USA) were incubated at 4 °C overnight, and the secondary antibodies FITC conjugated Goat Anti-Rabbit (GB22303, Servicebio, China) and Cy3 conjugated Goat Anti-mouse (GB21301, Servicebio, China) were incubated at 37 °C for 30 min. Finally, after incubation of DAPI (G1012, Servicebio, China) at room temperature for 10 min, the anti-fluorescence attenuation tablets were sealed. Fluorescence microscope (OlyVIA, OLYMPUS, Japan) was used for observation and photography.

Mouse brain tissues were collected and total RNA was extracted using Molpure Cell/Tissue total RNA kit (19221ES50, YEASEN, China), and total RNA was reversed into cDNA using PrimeScript RT reagent kit (RR047A, Takara, China). TB Green Premix Ex Taq Ⅱ (Tli RNaseH Plus) (RR820A, Takara, China) was used to perform a PCR reaction with primers of each gene. The specific primer sequences were shown in Table 1. The relative mRNA expression levels of each gene were calculated by 2^−△△Ct with β-actin as the internal reference.

WB was used to detect the expression of the short-chain fatty acid (SCFA) receptor Olfr78 in the hippocampus. The protein lysate was obtained from the hippocampus by RIPA (P0013, Beyotime, China), quantitated and denatured, and then separated by SDS-PAGE and transferred to the PVDF membrane. After sealing with 5% skim milk, the primary antibody Olfr78 (PA5-75357, Invitrogen, USA) and β-actin (AC026, abclonal, China) were incubated at 4℃ overnight, and the secondary antibody (S0001, Affinity biosciences, China) was incubated at room temperature for 2 h. After the reaction, the color reaction was carried out by ECL (17046, zen-bio, China) and Tanon fluorescence image analysis system (5200 Multi, Tanon, China), and the relative protein expression of Olfr78 was calculated.

Colonic contents were collected aseptically, total DNA was extracted, quantified and measured by Nanodrop and 1.2% agar-gel electrophoresis. The target sequences that can reflect the composition and diversity of bacterial flora, such as microbial ribosomal RNA or specific gene fragments, were taken as targets, corresponding primers were designed according to the conserved regions in the sequences, and specific Barcode sequences were added to the samples, and then PCR amplification was performed on the variable regions of rRNA genes or specific gene fragments. The amplified products were purified by magnetic beads, and fluorescence quantification was performed using a quant-iT picoGreen dsDNA assay kit (P7589, Invitrogen, USA) and a microplate reader (FLx800, BioTek, USA). According to the fluorescence quantitative results, each sample was mixed according to the corresponding proportion according to the sequencing quantity requirement of each sample. TruSeq Nano DNA LT Library Prep Kit (Illumina, USA) was used to prepare sequencing libraries and perform high-throughput sequencing. The original data were collected and the results were analyzed.

Acetic acid, propionic acid, butyric acid, isobutyric acid, valeric acid, isovaleric acid, and caproic acid were used to prepare the standard stock solutions. In addition, blood samples of mice in each group were collected and centrifuged at 3000 rpm for 15 min. Appropriate serum was collected and added with 50 µL 15% phosphoric acid, 100 µL 125 µg/mL internal standard solution (isocaproic acid), and 400 µL ether, mixed for 1 min, centrifuged at 12,000 rpm at 4 ℃ for 10 min, and supernatant was obtained for detection. The Gas chromatography analysis was performed under the trace 1300 gas chromatograph (Thermo Fisher Scientific, USA) with Agilent HP-INNOWAX capillary column (30 m × 0.25 mm ID × 0.25 μm) and helium was used as the carrier gas at 1 mL/min. The sample size was 1 µL and the shunt ratio was 10:1. The injector temperature is 250℃, the ion source temperature is 300℃, and the transmission Line temperature is 250℃. The column temperature was programmed to increase from an initial temperature of 90 ℃, followed by an increase to 120 ℃ at 10 ℃/min, and to 150 ℃ at 5 ℃/min, and finally to 250 ℃ at 25 ℃/min which was Maintained for 2 min. Mass spectrometry was performed by ISQ 7000 mass spectrometer (Thermo Fisher Scientific, USA) with electron impact ionization mode. Single ion monitoring (SIM) mode was used with an electron energy of 70 eV.

All experimental data were analyzed using SPSS software, and finally expressed as mean ± standard deviation. All data were normally distributed. Independent sample t-test was used for differential expression between the two groups, and one-way analysis of variance was used for differential expression between the multiple groups. P < 0.05 was considered to have a significant difference in statistical results.

During the experiment, the weight of the mice was monitored every week. As shown in Fig. 2A, the weight of mice in the T2DM/Pg + T2DM group decreased significantly from the 8th week (P < 0.001). There was no difference in body weight between Pg and control group (P > 0.05), but the body weight of mice in the Pg + T2DM group also decreased significantly from the 8th week compared with the T2DM group (P < 0.05), indicating that Pg only affected body weight of T2DM mice. In addition, blood glucose monitoring of mice showed that the blood glucose of mice in the T2DM/Pg + T2DM group was significantly higher than that in the control/Pg group (P < 0.001, Fig. 2D). Pg did not affect the random blood glucose level of control mice (P > 0.05), but significantly reduced the random blood glucose level of mice in the T2DM group from the 8th week (P < 0.01, Fig. 2D). Similarly, the OGTT results showed that the blood glucose levels of the model mice significantly increased after glucose intake, and the decline rate was slow (Fig. 2B). Even after 120 min, the levels were still higher than the normal level, with a significant increase in AUC (P < 0.001, Fig. 2C), indicating impaired glucose tolerance. Pg significantly reduced the AUC of the model mice, and the glucose tolerance of the model mice improved (P < 0.001, Fig. 2B and C). The ITT results also showed that the blood glucose levels of the model mice decreased slowly after insulin stimulation, with a significant increase in AUC, and insulin sensitivity decreased (P < 0.001, Fig. 2E and F). Pg significantly reduced the AUC of the model mice and enhanced their insulin sensitivity (P < 0.001, Fig. 2E and F). Compared with the control/Pg group, the levels of TC, TG, and LDL-C in the T2DM/Pg + T2DM group were significantly increased, and the levels of HDL-C were significantly decreased (P < 0.001, Fig. 2G). Pg did not affect lipid levels in control mice (P > 0.05, Fig. 2G), but reduced LDL-C levels in the T2DM mice (P < 0.01, Fig. 2G). Overall, Pg did not affect normal mice and only had a partial effect on glycolipid metabolism in T2DM mice.

Further behavioral tests of the mice showed no difference in average swimming speed between the groups during the water maze experiment (P > 0.05, Fig. 2H). In the test stage, the escape latency of mice in the T2DM/Pg + T2DM group was significantly increased compared with the control/Pg group (P < 0.05, Fig. 2H), and the number of times crossing the platform was also significantly reduced (P < 0.01, Fig. 2H), indicating that the spatial learning ability of mice was impaired by T2DM modeling. Pg treatment further increased the escape latency of mice in the T2DM group and reduced the number of mice passing through the platform (P < 0.05, Fig. 2H), indicating that Pg was involved in the spatial learning impairment caused by T2DM. In addition, the results of the new object recognition experiment showed that the NOI of mice in the T2DM/Pg + T2DM group was significantly lower than that in the control/Pg group (P < 0.001, Fig. 2I). Pg had no effect on the NOI of normal mice (P > 0.05), but it would further reduce the NOI of T2DM mice (P < 0.01, Fig. 2I). Water maze and new object recognition experiments confirmed the cognitive impairment in T2DM mice, and Pg could further aggravate the cognitive impairment caused by T2DM.

Further analysis of the damage of mice hippocampal neurons showed that dark-colored neurons were observed in the hippocampal region of brain tissue of T2DM mice, and pathological changes such as vacuolar degeneration, necrosis, and loose arrangement of vertebral cells were observed (Fig. 3A). Compared with T2DM mice, the lesions of the Pg + T2DM group were more severe (Fig. 3A). Toluidine blue staining results showed that compared with the control group, the number of Nissl bodies in the hippocampus of mice in T2DM and Pg groups was significantly decreased (P < 0.001), and the number of Nissl bodies in the Pg + T2DM group was further reduced compared with the T2DM group (P < 0.001, Fig. 3B and C). In addition, TUNEL staining showed that the percentage of apoptotic cells in the hippocampal tissue of mice in both T2DM and Pg groups was significantly increased (P < 0.05), and the percentage of apoptotic cells in the Pg + T2DM group was further increased than that in the T2DM group (P < 0.001, Fig. 3D and E). These results suggested that the hippocampal neurons of T2DM mice have obvious damage, and Pg aggravates the damage caused by T2DM.

The ultrastructure of hippocampal neurons and synapses was observed by transmission electron microscopy (Fig. 4A), and the width and length of PSD were measured (Fig. 4B). Results showed that compared with the control group, the width and length of hippocampal PSD in the T2DM and Pg groups were significantly reduced (P < 0.001), and the PSD width and length in the T2DM/Pg group were further reduced in the Pg + T2DM group (P < 0.001, Fig. 4B). Immunofluorescence analysis of SYN expression also showed that SYN expression was significantly decreased in the T2DM and Pg groups, and Pg + T2DM further decreased SYN expression in the T2DM/Pg groups (P < 0.05, Fig. 4C and D). Further detection of related genes by RT-PCR showed that the expressions of PSD-95 (Fig. 4E), fragile X-related 1 (FXR1, Fig. 4F), FXR2 (Fig. 4G), GluN2B (Fig. 4I) and GluA2 (Fig. 4J) in the brain tissue of mice in T2DM and Pg groups were significantly decreased (P < 0.05), and Pg + T2DM further decreased the expression of these genes in the brain tissue of mice in the T2DM/Pg group (P < 0.05). These results suggested that hippocampal synaptic plasticity was significantly decreased in both T2DM and Pg groups, and Pg + T2DM could further reduce synaptic plasticity in T2DM mice.

The effects of Pg on the inflammatory response of microglia in T2DM mice were further analyzed. The expression of neuroinflammatory factors in brain tissue was examined by RT-PCR, and it was found that the expressions of IL-1β, iNOS, and COX2 in brain tissue of T2DM mice were significantly increased (P < 0.05), and the expression of inflammatory factors in brain tissue of Pg + T2DM group was further increased compared with that of T2DM/Pg group (P < 0.05, Fig. 5A-C). Immunofluorescence detection of Iba1 was performed to analyze the activation of microglia. It was found that the expression of Iba1 in the hippocampus of mice in both T2DM and Pg groups was significantly increased, and the expression of Iba1 was further increased in the Pg + T2DM group (P < 0.001, Fig. 5D and E). The expressions of microglia-neuron interaction genes BDNF (Fig. 5F), NGF (Fig. 5G), DAP12 (Fig. 5H), CX3CR1 (Fig. 5I) and CX3CL1 (Fig. 5J) showed that the expression of BDNF in the T2DM group was significantly decreased (P < 0.05), and Pg had no significant regulatory effect on BDNF in the control group (P > 0.05, Fig. 5F). And the expression of NGF was significantly decreased in both T2DM and PG groups, while the expressions of DAP12, CX3CR1, and CX3CL1 were significantly increased (P < 0.05, Fig. 5F-J). The Pg + T2DM group further regulates the expression of these genes in the T2DM and Pg groups (P < 0.05, Fig. 5F-J). These results indicated that Pg significantly promoted the activation of hippocampal microglial inflammatory response induced by T2DM.

Studies have shown that the gut microbiota of T2DM patients is disordered, and the disordered gut microbiota can affect cognitive levels through the gut-brain axis. Pg may also further cause gut microbiota disturbance in T2DM, thereby mediating the gut-brain axis effect on cognition. We analyzed the gut microbiota of mice in each group through 16 s rRNA sequencing, as shown in Fig. 6. The analysis results of α diversity were shown in Fig. 6A, richness was represented by the Chao1 index, evolution-based diversity was represented by Faith's PD index, and coverage was represented by Good's coverage index. And Jaccard-based PcoA was used to demonstrate beta diversity (Fig. 6B). Non-dimensional multidimensional scaling analysis (NMDS) also simplifies the data structure through dimensionality reduction decomposition of the sample distance matrix, so as to describe the distribution characteristics of samples at a specific distance scale. Jaccard-based NMDS was shown in Fig. 6C. Both α and β diversity indicated that there were some differences in microbial communities among samples of each group. To investigate which species are common and which are unique among different groups, community analysis was performed using the Venn diagram, as shown in Fig. 6D. The species composition differences among samples were further compared, and the clustering tree ranking was drawn according to the correlation between samples, that is, the clustering heat map was drawn, as shown in Fig. 6E and F. In addition, the similarity between samples is displayed in the form of hierarchical tree, and the clustering effect is measured by the branch length of the clustering tree, the Jaccard-based clustering tree as shown in Fig. 6G. LEfSe analysis is an analytical method that combines non-parametric Kruskal-Wallis and Wilcoxon rank sum tests with linear discriminant analysis effect size, which can directly analyze the difference of all classification levels at the same time. The histogram of the LDA value distribution of different species in this study was shown in Fig. 6H.

In addition, the serum metabolomics was further analyzed by targeting SCFAs. The sample level clustering tree was shown in Fig. 7A, and the overall metabolite clustering heat map was shown in Fig. 7B. Analysis results of specific differential metabolites in serum metabolomics were shown in Fig. 7C-E, mainly involving acetic acid, propionic acid, and isobutyric acid. A further analysis of the correlation between SCFAs and the intestinal microbiota was conducted, and the results are shown in Fig. 7F. Acetic acid, isobutyric acid, and isovaleric acid have significant correlations with various gut microbiota, including unidentified_S24-7, Lactobacillus, Odoribacter, [Ruminococcus], unclassified_Clostridiales, unidentified_Desulfovibrionaceae, unidentified_Peptostreptococcaceae, Allobaculum, and Bifidobacterium, etc. In addition, the expression of the SCFA receptor Olfr78 in the hippocampus of mice was further analyzed by immunofluorescence. The results showed that both T2DM and Pg significantly reduced the expression of Olfr78 in the hippocampus of control mice, and Pg could further reduce the expression of Olfr78 in the hippocampus of T2DM mice (P < 0.01, Fig. 8A and B). The results of WB were similar to those of immunofluorescence (Fig. 8C and D), suggesting that Pg plays a role in the regulation of the SCFA metabolism pathway in T2DM.