Inflammatory Manifestations Associated With Gut Dysbiosis in Alzheimer's Disease

Abstract

Recent studies strongly suggest that gut microbiome can influence brain functions and contribute to the development of Alzheimer's disease (AD). However, reported changes in the gut microbiomes in AD patients from different countries are not similar, and more research is needed to reveal the relationships between human microbiomes and AD in diverse ethnic populations. There is also an assumption that microbiome-associated peripheral inflammation might drive the development of sporadic AD. This cross-sectional study is aimed at analyzing the gut microbial profile and exploring potential associations with blood cytokines and some clinical parameters among individuals diagnosed with Alzheimer's in Kazakhstan. Consistent with previous studies, we have found that the microbial landscape in AD reveals specific alterations in the gut microbiome. Specifically, the AD patient group showed a decreased Firmicutes/Bacteroidetes ratio. The differential abundance analysis highlighted a dysbiosis in the gut microbiota of AD patients, marked by a reduced presence of Bifidobacterium, particularly B. breve. In our study, AD patients' altered gut microbiota composition notably features an increased presence of Pseudomonadota like Phyllobacterium and inflammatory bacteria such as Synergistetes and the Christensenellaceae family. The metabolic profiling of the AD microbiome reveals a predominant presence of pathways related to sugar, carrier molecules, tetrapyrrole, pyrimidine biosynthesis, and nucleic acid processing. This analysis also highlighted a marked reduction in SCFA, carbohydrate, polysaccharide, polyamine, and myo-inositol degradation pathways. The increases in the proinflammatory cytokines IL-1a, IL-8, IL-17A, IL-12p40, TNF-β, MCP-1, IL-2, and IL-12p70 and the anti-inflammatory cytokines IL-10 and IL-13 were observed in AD patients. Key variables driving the separation of AD and controls include inflammatory markers (IL-1a and IL-8), growth factors (EGF), lipids (LDL), BMI, and gut microbes, like genus Tyzzerella and Turicibacter and species Parabacteroides distasonis and Bacteroides eggerthii. We have also demonstrated that almost all cytokines strongly correlate with serum adiponectin levels and specific microbial taxa in AD patients. Thus, our findings identify potential microbial and inflammatory signatures in an ethnically distinct cohort of AD patients. These could serve as AD biomarkers and microbiota-based therapeutic targets for treating AD.

Keywords: Alzheimer's disease; cytokines; dysbiosis; gut microbiome; serum adiponectin.

Figure 1

Bacterial community composition in the gut of patients with Alzheimer's disease (AD) compared to controls: (a–c) within-sample diversity of the gut microbiome in AD patients and controls (Mann–Whitney U test); (d) boxplot showing the ratio of Firmicutes to Bacteroidetes (F/B) of bacterial phyla in AD and control samples (Mann–Whitney U test); (e) bar plot of log2-fold change in abundance (L2FC) indicating differentially enriched bacterial taxa in the AD group (purple) compared to the control group (green) (calculated using DESeq2 (p < 0.05, FDR), plotted using LEfSe); (f) ordination using principal coordinate analysis (PCoA) based on Bray–Curtis dissimilarity showing differences in gut microbial composition between groups (beta diversity), (ANOSIM, R = 0.03, p = 0.0466; PERMANOVA, F = 2.01, p = 0.036); (g) bacterial genera overdepleted in the AD group (Mann–Whitney U test); (h) the cladogram represents differentially abundant bacterial taxa; the center of the cladogram represents the kingdom; each subsequent circle is phylogenetically one level lower (phylum, class, order, family, genus, and species). Areas in purple represent taxa enriched in the AD group, and green areas are enriched in the control group. Calculated using DESeq2 (p < 0.05, FDR) and plotted using the LEfSe cladogram algorithm. p, phylum; c, class; o, order; f, family; g, genus; s, species.

Figure 2

Extended error bar plots showing functional properties that differ between control and Alzheimer's gut microbiomes (T-test or Mann–Whitney U test, FDR, p ≤ 0.05, no 90% CI overlap for difference between medians). The left side shows the relative abundance of metabolic features based on gut microbiome abundance, and the right part visualizes the difference in median abundances between the groups for each feature. Purple indicates the Alzheimer's group; green is the control group. Independent T-test, Welch's T-test, or Mann–Whitney U-test, where appropriate. FDR, p ≤ 0.05. 90% confidence intervals (CI) for differences between medians constructed using the Hodges–Lehmann method. The relative abundance is scaled (multiplied by 100).

Figure 3

Cytokine expression profiles in Alzheimer's disease patients (AD) and controls: (a) cytokine levels differ between the AD group (purple) and the control (green) (Mann–Whitney U test, FDR, p ≤ 0.05); (b, c) correlation analyses between clinical variables and immunological parameters and significantly differentially abundant taxa identified in DESeq2 analysis and immunological parameters. On correlation cluster-grams - parameters enriched in AD are highlighted in purple, and control is highlighted in green on the sides of a heatmap. Red indicates a positive correlation; blue indicates a negative correlation. Spearman r, FDR, 0.05. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001.

Figure 4

Feature importance analysis for discriminating between Alzheimer's disease (AD) patients and controls: (a) the importance bar plot shows the 20 most important blood inflammatory markers, the top discriminators between the AD and control groups; (b) the importance bar plot shows the top 20 discriminators between the AD and control groups; the horizontal axis represents the measured effect of the feature on discrimination (“importance”). Feature names ranked by importance (↑) are shown on the vertical axis. (c) principal component analysis (PCA) ordination based on the most important z score normalized discriminative features. PCA loadings are shown as arrows; the length of the arrow indicates the relative effect of the feature on the ordination, and its angle indicates the direction of the impact. Feature importance analysis was performed using the gradient boosting decision trees (GBDT) algorithm with performance evaluation using leave-one-out (LOO) cross-validation with the area under the precision-recall curve (AUC) as the performance metric. The significance of performance based on identified important features (IF) was assessed using a permutation importance test with 9999 permutations. ADIPOQ was excluded from the analysis.