Transmission of Alzheimer's Disease-Associated Microbiota Dysbiosis and its Impact on Cognitive Function: Evidence from Mouse Models and Human Patients

Abstract

Spouses of Alzheimer's disease (AD) patients are at higher risk of developing AD dementia, but the reasons and underlying mechanism are unknown. One potential factor is gut microbiota dysbiosis, which has been associated with AD. However, it remains unclear whether the gut microbiota dysbiosis can be transmitted to non-AD individuals and contribute to the development of AD pathogenesis and cognitive impairment. The present study found that co-housing wild-type mice with AD transgenic mice or giving them AD transgenic mice feces caused AD-associated gut microbiota dysbiosis, Tau phosphorylation, and cognitive impairment. Gavage with Lactobacillus and Bifidobacterium restored these changes. The oral and gut microbiota of AD patient partners resembled that of AD patients but differed from healthy controls, indicating the transmission of oral and gut microbiota and its impact on cognitive function. The underlying mechanism of these findings includes that the butyric acid-mediated acetylation of GSK3β at lysine 15 regulated its phosphorylation at serine 9, consequently impacting Tau phosphorylation. These results provide insight into a potential link between gut microbiota dysbiosis and AD and underscore the need for further research in this area.

Figure 1. WT mice developed cognitive impairment after co-housing with AD Tg mice.

a. The 2-months-old WT mice co-housed with 2-months-old AD Tg mice for up to 3 months are defined as ADWT mice. The ADWT mice were separated from AD Tg mice at age of 5-months-old. The behavioral tests of mice were performed at age of 3, 5 and 8 months old. After co-housing with AD Tg mice for 3 months, the AD Tg mice and ADWT mice developed cognitive impairment compared to the WT mice, as demonstrated in increased MWM escape latency during training days (b), decreased MWM platform crossing numbers on testing day (c), but no significant changes in swimming speed in MWM (d), increased latency to enter escape box during BM training days (e), and increased latency to enter escape box on BM testing day (f), decreased BM target time on testing day (g), increased number of wrong holes searched on BM testing day (h), increased distance on testing day (i), but no significant changes in speed during training days (j) compared to the WT mice. Data are mean ± standard deviation or medians (with interquartile ranges), N = 12–14 mice in each experimental group. The P values refer to the differences of variables among the groups, * P < 0.05; ## P < 0.01. Two-way ANOVA with repeated measurement and Bonferroni correction was used to analyze the data presented in b, d, e, and j. The P values refer to the interaction of group in MWM and BM training days. One-way ANOVA with Bonferroni correction was used to analyze the data presented in c, f, g, h, and i. AD, Alzheimer’s disease; WT, wild-type; ADWT, AD-exposed WT; Tg, transgenic; MWM, Morris water maze; BM, Barnes maze.

Figure 2. WT mice developed cognitive impairment after fecal microbiota transplantation from AD Tg mice.

a. The 2-months-old WT mice received gavage of fecal microbiota from the 2-months-old WT or AD Tg mice for 7 days; the behaviors of the recipient WT mice were tested one month after the gavage at 3-months-old. The recipient WT mice received AD mice fecal microbiota transplantation developed cognitive impairment compared to the WT mice received saline, as demonstrated as increased escape latency during training days (b), decreased platform crossing number on testing day (c), but no significant changes in swimming speed (d) of MWM. The recipient WT mice received AD mice fecal microbiota transplantation developed cognitive impairment compared to the WT mice received saline, as demonstrated in increased latency to enter escape box during BM training days (e), increased latency to enter escape box on BM testing day (f), increased number of wrong holes searched on BM testing day (h), but not significant changes in BM target time on testing day (g), no significant changes on distance on BM testing day (i), and no significant changes on speed during training days. N = 9 mice in each experimental group. Two-way ANOVA with repeated measurement and Bonferroni correction was used to analyze the data presented in b, d, e, and j. The P values refer to the interaction of group in MWM and BM training days. One-way ANOVA with Bonferroni correction was used to analyze the data presented in c, f, g, h, and i. The P values refer to the differences of variables between the groups, * P < 0.05. AD, Alzheimer’s disease; WT, wild-type; Tg, transgenic; MWM, Morris water maze; BM, Barnes maze; FMT, fecal microbiota transplantation.

Figure 3. ADWT mice acquired AD-associated gut microbiota dysbiosis after co-housing with AD Tg mice.

a. Experimental design of fecal collection after 3 months co-housing. b. Principal component analysis (PCA) using the Bray–Curtis dissimilarity metric among fecal samples of WT, ADWT, and AD Tg mice (P = 3.56e-9; permutational multivariate analysis of variance, PERMANOVA). Each dot represents an individual. PC1, PC2, and PC3 represent the percentage of variance explained by each coordinate. c. The Simpson diversity of ADWT mice was similar to that of AD Tg mice, but different from that of WT mice. d. Heatmap indicated the different changes in bacterial community structure represented as relative abundance shown in genus level analysis among the AD Tg, ADWT and WT mice. Relative to the WT mice, the AD Tg and ADWT mice had higher relative abundance of Peptococcaceae(e); Butyricoccuss (f); Dubosiella (g); Herbinix (h); Coribacteriales (i); Romboutsia (j), but lower relative abundance of Marvinbryantia (l); Bacteroides (m); Faecalibaculum (n); Methanosarcina (o); Muribaculum (p); Lactobacillus (q); Alistipes (r); Ruminclostridium_1 (s). Relative to the WT mice, the AD Tg mice had higher relative abundance of Ruminclostridium_5 (k) but lower relative abundance of Lachnoclostridium (t). u. The heat map demonstrated the pair-wise correlative relationship between any pair two bacteria among the WT, ADWT, and AD Tg mice. h. Changes in fecal short chain fat acids showed that the AD Tg and ADWT mice had less butyric acid compared to the WT mice had. N = 8–12 biologically independent samples in each group. * P < 0.05; ** P < 0.01. MaAsLin 2 implementation were used for testing in microbiome profiles taxonomic result (e to t). AD, Alzheimer’s disease; WT, wild-type; PC, principal component.

Figure 4. Differences in brain levels of SCFAs, phosphorylated Tau, IL-6 and Ab among AD Tg, ADWT and WT mice.

a. The AD Tg mice had less brain acetic acid, but not propionic acid, levels compared to WT mice. Both the AD Tg and ADWT mice had less butyric acid levels in brain tissues compared to the WT mice. b. Western blot shows that the AD Tg and ADWT mice had higher amounts of Tau-pS202/PT205 and Tau-pS262, but not total Tau, in the hippocampus compared to WT mice. c. The quantification of the Western blots showed that the AD Tg and ADWT mice had a higher ratio of Tau-pS202/PT205 to total Tau and Tau-pS262 to total Tau in the hippocampus compared to the WT mice. d. ELISA showed that AD Tg and ADWT mice had higher Tau-pS199 amounts in the hippocampus compared to WT mice. e. The AD Tg and ADWT mice had higher amounts of IL-6 in the hippocampus compared to WT mice. f. The AD Tg and ADWT mice had higher amounts of Ab42 and Ab40 in the hippocampus compared to the WT mice. N = 3 – 8 biologically independent samples in each group. One-way ANOVA with Bonferroni correction was used to analyze the data presented in a, c, d, e, f and g. * P < 0.05; ** P < 0.01. AD, Alzheimer’s disease; WT, wild-type; Tg, transgenic; Interleukin 6, IL-6.

Figure 5. Butyric acid increased GSK3b-S9 levels dependent on acetylation of lysine at 15.

a. Western blot showed that the AD Tg and ADWT mice had higher amounts of Tau-pS202/pT205 and lower amounts of GSK3b-S9 in the hippocampus compared to WT mice. b. The quantification of the Western blot demonstrated that AD Tg and ADWT mice had higher ratio of Tau-PS202/PT205 to total Tau in the hippocampus compared to WT mice. c. The quantification of the Western blot demonstrated AD Tg and ADWT mice had lower ratio of p-GSK3b-S9 to total GSK3b in the hippocampus compared to WT mice. d. Western blot showed that the butyric acid induced a dose-dependent increase in p-GSK-3b-S9 levels in HEK293T cells and LY2090314 (LY), the inhibitor of GSK-3, blocked the effect of butyric acid. e. The quantification of the Western blot demonstrated the dose-dependent effects of butyric acid on increasing the ratios of p-GSK-3b-Ser9 to GSK3bin HEK293T cells, LY block the effect of butyric acid. f. Annotation of representative tandem mass spectra of Trypsin-GluC digested GSK3b, depicting K15 acetylation following the treatment of butyric acid. g. The computer-generated WT and 3 independent site-directed mutations (K15R, K15R/A11K, and K15R/S13K). h. The effects of butyric acid on amounts of p-GSK3b-Ser9 and GSK3b in WT and the 3 independent site-directed mutants (K15R, K15R/A11K, and K15R/S13K) HEK293T cells. i. The 3 mutations did not significantly change the base line ratios of p-GSK-3b-Ser9 to GSK3b. j. However, the mutations of K15R increased the ratios of p-GSK-3b-Ser9 to GSK3b; the mutations of K15R plus A11K had greater, but the mutations of K15R plus A13S had lesser, increases in the ratios of p-GSK-3b-Ser9 to GSK3bthan the mutations of K15R following the butyric acid treatment. j. The hypothesized pathway showing that lysine at 15 of GSK3b may play an important role in the butyric acid-mediated inhibition of GSK3bactivity, Tau phosphorylation and cognitive impairment. N = 3 biologically independent samples in each group. AD, Alzheimer’s disease; WT, wild-type; p, phosphorylated; PS, phosphorylated serine; PT, phosphorylated threonine; LY2090314, LY; K, lysine; R, arginine; S, serine; A, alanine.

Figure 6. Treatment with bacteria (Lactobacillus and Bifidobacterium) mitigated the behavioral and cellular changes in ADWT mice.

a. The gavage of Lactobacillusplus Bifidobacterium (in the first 10 days in each month of the total 3 months) increased fecal butyric acid amounts of the ADWT mice compared to saline treatment. The gavage of Lactobacillus plus Bifidobacterium reduced the amounts of Tau-pS202/pT205 and Tau-pS199 (b), IL-6 (c) and Ab42 and Ab40 (d) in the hippocampus of the ADWT mice compared to saline treatment. Finally, the ADWT mice with gavage of Lactobacillus plus Bifidobacterium had better cognitive function compared to the ADWT mice with saline treatment, as demonstrated in MWM training (e), MWM testing (f), BM training (g), and BM testing (h). i, The hypothesized pathway showing that ADWT mice, resulting from the co-housing of AD Tg and WT mice, acquire the AD-associated gut microbiota dysbiosis, which causes reductions in gut and brain butyric acid amounts, leading toTau phosphorylation, IL-6 elevation and Ab accumulation, leading to the cognitive impairment in the ADWT mice. Data are mean ± standard deviation or median (with interquartile range), N = 6 to 15 biologically independent samples in each group. Student’s t-tests were used to analyze the data in a, b, c, and d. Two-way ANOVA with repeated measurement and Bonferroni correction was used to analyze the data presented in e and g. Mann Whitney U test was used to analyze the data in f and h. AD, Alzheimer’s disease; WT, wild-type; p, phosphorylated; pS, phosphorylated serine; pT, phosphorylated threonine; Interleukin 6, IL-6; MWM, Morris water maze; BM, Barnes maze.

Figure 7. Microbiota in oral and fecal samples of Alzheimer’s disease (AD) patients, partners of AD patients (PAD), and control (CON) individuals.

a. Schema of oral and fecal sample collection from AD, PAD, and CON. b. The average taxonomic distribution of bacteria from oral 16S RNA sequencing at the order level among the AD, PAD, and CON. c. Relative abundances of 9 bacteria in oral samples from AD and PAD were significantly lower than those in CON. d. The average taxonomic distribution of bacteria from fecal 16S RNA sequencing at the order level among the AD, PAD, and CON. e. Relative abundances of 3 bacteria in fecal samples from AD and PAD were significantly lower than those in CON. f. The ratios of acetic acid and butyric acid to total SCFAs in fecal samples among the three cohorts showed that the AD and PAD had higher acetic acid, but lower butyric acid compared to CON. There were the following biologically independent samples in each group of oral samples: AD (N = 19), PAD (N = 11), CON (N = 24) and fecal samples: AD (N = 39), PAD (N = 22), CON (N = 33). * or # P < 0.05; ** P < 0.01. MaAsLin 2 implementation were used for testing in microbiome profiles taxonomic result (c and e). One-way ANOVA with Bonferroni correction (f). AD, Alzheimer’s disease; PAD, partners of AD; and CON, control.