A peripheral signature of Alzheimer's disease featuring microbiota-gut-brain axis markers

Abstract

Background: Increasing evidence links the gut microbiota (GM) to Alzheimer's disease (AD) but the mechanisms through which gut bacteria influence the brain are still unclear. This study tests the hypothesis that GM and mediators of the microbiota-gut-brain axis (MGBA) are associated with the amyloid cascade in sporadic AD.

Methods: We included 34 patients with cognitive impairment due to AD (CI-AD), 37 patients with cognitive impairment not due to AD (CI-NAD), and 13 cognitively unimpaired persons (CU). We studied the following systems: (1) fecal GM, with 16S rRNA sequencing; (2) a panel of putative MGBA mediators in the blood including immune and endothelial markers as bacterial products (i.e., lipopolysaccharide, LPS), cell adhesion molecules (CAMs) indicative of endothelial dysfunction (VCAM-1, PECAM-1), vascular changes (P-, E-Selectin), and upregulated after infections (NCAM, ICAM-1), as well as pro- (IL1β, IL6, TNFα, IL18) and anti- (IL10) inflammatory cytokines; (3) the amyloid cascade with amyloid PET, plasma phosphorylated tau (pTau-181, for tau pathology), neurofilament light chain (NfL, for neurodegeneration), and global cognition measured using MMSE and ADAScog. We performed 3-group comparisons of markers in the 3 systems and calculated correlation matrices for the pooled group of CI-AD and CU as well as CI-NAD and CU. Patterns of associations based on Spearman's rho were used to validate the study hypothesis.

Results: CI-AD were characterized by (1) higher abundance of Clostridia_UCG-014 and decreased abundance of Moryella and Blautia (p < .04); (2) elevated levels of LPS (p < .03), upregulation of CAMs, Il1β, IL6, and TNFα, and downregulation of IL10 (p < .05); (3) increased brain amyloid, plasma pTau-181, and NfL (p < 0.004) compared with the other groups. CI-NAD showed (1) higher abundance of [Eubacterium] coprostanoligenes group and Collinsella and decreased abundance of Lachnospiraceae_ND3007_group, [Ruminococcus]_gnavus_group and Oscillibacter (p < .03); (2) upregulation of PECAM-1 and TNFα (p < .03); (4) increased plasma levels of NfL (p < .02) compared with CU. Different GM genera were associated with immune and endothelial markers in both CI-NAD and CI-AD but these mediators were widely related to amyloid cascade markers only in CI-AD.

Conclusions: Specific bacterial genera are associated with immune and endothelial MGBA mediators, and these are associated with amyloid cascade markers in sporadic AD. The physiological mechanisms linking the GM to the amyloid cascade should be further investigated to elucidate their potential therapeutic implications.

Keywords: Alzheimer’s disease; Cognitive impairment; Endothelial dysfunction; Gut microbiota; Lipopolysaccharide; Microbiota-gut-brain axis.

Fig. 1

Gut microbial communities of study participants grouped at phylum (A) and genus (B) levels. P-values were calculated using the online LEfSe workflow on the Hutlab Galaxy platform. Log10 transformation of genera abundances was used to increase readability. Abbreviations: CU, cognitively unimpaired persons; CI-NAD, patients with cognitive impairment not due to AD; CI-AD, patients with cognitive impairment due to AD

Fig. 2

Blood microbiota-gut-brain axis mediators in the blood of study participants. P-values were calculated by using one-way ANOVA with Bonferroni’s correction for continuous Gaussian variables (or Kruskall-Wallis test with Dunn’s correction for non-Gaussian variables) for the Gram-negative membrane protein LPS (A), the pro (IL1β, IL6, TNFα, IL18) and anti-inflammatory (IL10) cytokines (B) as well as the soluble cell adhesion molecules (C). Data are presented as box-plots with black horizontal lines indicating medians and circles indicating the subject’s values. Abbreviations: CU, cognitively unimpaired persons; CI-NAD, patients with cognitive impairment not due to AD; CI-AD, patients with cognitive impairment due to AD

Fig. 3

Markers of the amyloid cascade of study participants. P-values were calculated by using one-way ANOVA with Bonferroni’s correction for continuous Gaussian variables (or Kruskall-Wallis test with Dunn’s correction for non-Gaussian variables) for the amyloid load (A), plasma NfL (C) and cognitive measures (D–E) but not for pTau-181 (B), where Mann–Whitney test was applied. Data are presented as box-plots with black horizontal lines indicating medians and circles indicating the subject’s values. Abbreviations: CU, cognitively unimpaired persons; CI-NAD, patients with cognitive impairment not due to AD; CI-AD, patients with cognitive impairment due to AD

Fig. 4

Association matrices of genera with MGBA mediators (A) and of the latter with amyloid cascade markers (B). Heatmap of the Spearman’s rho coefficient values (pink: positive; green: negative) indicating significant association (age adjusted, p < 0.05) in CU and CI-NAD or CU and CI-AD. The star indicates moderate associations (Spearman’s rho value > 0.4). For MMSE and ADAS-cog, higher values reflected higher cognitive scores. Abbreviations: CU, cognitively unimpaired persons; CI-NAD, patients with cognitive impairment not due to AD; CI-AD, patients with cognitive impairment due to AD

Fig. 5

Closest associations (Spearman’s rho value > 0.4) between fecal bacterial genera and microbiota-gut-brain axis mediators and between the latter and amyloid cascade markers in CU and CI-NAD (A) and CU and CI-AD (B)