The Gut Microbiota-Brain Axis during Aging, Mild Cognitive Impairment and Dementia: Role of Tau Protein, β-Amyloid and LPS in Serum and Curli Protein in Stool

The study addresses how aging-related changes in the gut microbiota may contribute to mild cognitive impairment and dementia via systemic inflammation and amyloid/tau pathology. Dementia affects tens of millions worldwide and is projected to increase markedly, posing major health and economic challenges. Aging is associated with reduced gut microbial diversity, increased intestinal permeability, and chronic low-grade inflammation (inflammaging), factors hypothesized to influence cognitive decline. The authors investigate whether dysbiosis and bacterial components (LPS and the functional amyloid curli) relate to circulating tau and β-amyloid and cognitive status, exploring the gut–brain axis in humans and in an aging rat model.

Background cited includes: dementia prevalence and costs; inflammaging as a hallmark of aging; the microbiota–gut–brain axis influencing CNS function and cognition; Alzheimer’s disease accounting for 60–80% of dementias and characterized by β-amyloid plaques and tau tangles; LPS-induced neuroinflammation increasing Aβ generation; genetic susceptibility (ApoE ε4) as a risk factor; reports of reduced anti-inflammatory commensals such as Faecalibacterium prausnitzii in older adults; and microbial functional amyloids (e.g., curli) influencing host protein aggregation. These studies motivate examining microbial dysbiosis and bacterial products as potential contributors to neurodegeneration.

Human study: Cross-sectional design recruiting volunteers from central Mexico. Groups: 35 healthy young (BMI ≤ 25), and 87 ≥60 years subdivided into 42 aging without cognitive impairment (A), 32 with mild cognitive impairment (MCI), and 13 with dementia (D). Inclusion/exclusion criteria per Supplementary Materials. Assessments: medical exam, anthropometry (InBody 720 BIA for body composition; height with biospace BSM370), blood pressure (Omron HEM-781INT). Cognitive evaluation: Mini-Mental State Examination (MMSE). Fasting blood collection (12 h), serum stored at −80°C. Biochemistry: glucose, total cholesterol, HDL-C, LDL-C, triglycerides (COBAS C111). Branched-chain amino acids (Abcam Ab83374). ApoE genotyping from leukocyte DNA (Qiagen kit), TaqMan probes rs429358 and rs7412 on ABI Prism 7900 HT; Hardy–Weinberg equilibrium verified. Serum biomarkers: tau (Abcam ab210972 Human Tau SimpleStep ELISA), β-amyloid 1-42 (Abcam Ab12030), LPS (CEB526Ge ELISA). Stool collection: DNA extraction (QIAMP DNA stool mini kit); 16S rRNA V3–V4 amplification with Illumina adapter-containing primers; amplicon purification (Ampure XP), sizing (Qiaxel), indexing (Nextera XT v2), quantification (Qubit 3.0), pooled equimolar library, sequencing on Illumina MiSeq (v3, 600 cycles; 2×300 bp; 15 pM with 20% PhiX). Bioinformatics: taxonomic assignment to phylum/genus/species; diversity metrics; linear discriminant analysis for differential taxa; PICRUSt for predicted functional pathways. Curli assessment in human stool: PCR for csgA (major curlin subunit) using specified primers; Western blot of stool proteins (30 μg per lane) using anti-CsgA_ECO57 primary (1:3500) and HRP-conjugated anti-rabbit secondary (Abcam ab6721, 1:20,000); GAPDH (ab181602, 1:40,000) as loading control; detection with ChemiDoc Image Lab; triplicate independent blots; protein quantification by Bradford (Bio-Rad). Animal study: Male Wistar rats housed at 22°C, 12:12 h light–dark, ad libitum food and water. Sampling at 2, 12, 18, and 24 months; at 24 months euthanasia by cervical decapitation and brain collection. Measurements: fecal microbiota profiling (as above), fecal curli protein by Western blot, brain curli protein and markers of inflammation/oxidative stress (TNFα, SOD2, catalase, Nrf2) by Western blot; serum LPS ELISA. Ethics: Human and animal protocols approved by INCMNSZ committees. Statistics: Data as mean ± SD; one-way ANOVA with Tukey post hoc; Spearman correlations; significance p < 0.05; PCA on 32 variables (anthropometric, biochemical, cognitive, and significant bacterial species), with centering/scaling, eigen decomposition, projection onto up to 3 components; visualization in R (factoextra); correlation matrix visualization with corrplot.

- Anthropometry/biochemistry: Older groups (A, MCI, D) had higher BMI (+19.5% vs young). MCI and D had higher body fat (by 4.5% and 13.8% vs A and Y respectively) and increased visceral fat. Dementia group showed 34.9% lower skeletal muscle mass vs others. Glucose was higher in MCI (+35.5%) and D (+42.1%) vs young; type 2 diabetes prevalence: A 2.4%, MCI 25%, D 15.4%. Triglycerides higher in A (+73% vs Y). LDL-C increased 40.2% (A) and 30.3% (MCI) vs Y; HDL-C decreased in MCI and D (men −34%; women −31.5% and −49% vs Y). - Serum inflammatory/neurodegeneration markers: LPS markedly increased with cognitive impairment: Y 26.4 ng/mL; A 157 ng/mL (~5× vs Y); MCI 399 ng/mL (~14×); D 694 ng/mL (~25×). β-amyloid was highest in D: 68.5 ng/mL (2.7× vs A; 68× vs Y). Tau protein was highest in D: 1326 pg/mL (35× vs Y). In subjects with type 2 diabetes, tau was 927 pg/mL, higher than in MCI (623 pg/mL). - BCAA: Serum BCAA increased by 102% (MCI) and 229% (D) vs Y in humans; in healthy aging rats (18–24 months), BCAA were lower vs adult rats (12 months), consistent with metabolic disease association in humans. - Gut microbiota (humans): Alpha diversity (Shannon) decreased with age and further in MCI and D. Phylum changes: MCI increased Firmicutes; D increased Fusobacteria. Genus level: A, MCI, D had increased Prevotella and decreased Bacteroides; A increased Succinivibrio; MCI increased Paraprevotella; D increased Escherichia. Species: Prevotella copri increased in A, MCI, D. LDA: Y enriched in Faecalibacterium prausnitzii and Ruminococcus bromii; A/MCI/D enriched in P. copri; MCI enriched in Prevotella stercorea; D enriched in Escherichia coli and Coprococcus eutactus. - Predicted functions (PICRUSt): Increased pathways in MCI/D vs Y/A for LPS biosynthesis, peptidoglycan biosynthesis, BCAA biosynthesis, and tryptophan metabolism; decreased glutamatergic synapse pathway. These aligned with elevated serum LPS and BCAA. - Curli protein (humans): All dementia subjects tested positive for curli in feces by PCR/Western. Curli abundance correlated positively with E. coli (p < 0.001) and was higher in MCI and D (approximately 4× and 9× vs Y). Curli abundance correlated inversely with MMSE score (p < 0.001). All subjects carrying ApoE ε4 had curli present. - Rat aging model: At 24 months, phylum-level shifts included increased Tenericutes and decreased Deferribacteres/Verrucomicrobia; genus-level increases in Bacteroides, Dorea, Blautia, Escherichia; species-level increases in E. coli at 18 and 24 months and Clostridium perfringens at 24 months. Fecal curli increased at 24 months; brain curli increased 1.4× (18 months) and 3.8× (24 months) vs young. Serum LPS increased 11× (12 months), 30× (18 months), 45× (24 months) vs 2 months. Brain TNFα increased 1.7× (18 months) and 3.9× (24 months); SOD2 increased 1.9× and 3.4×; Nrf2 decreased by 44% and 70% at 18 and 24 months, indicating neuroinflammation and oxidative stress. - Multivariate analyses: PCA separated A, MCI, and D; D associated with higher serum β-amyloid, tau, LPS, and fecal curli; MCI associated with higher fat mass, weight, BMI, triglycerides; A associated with higher MMSE, HDL-C, and Faecalibacterium prausnitzii. Correlation matrix: β-amyloid, curli, tau, LPS, and BCAA positively correlated with each other and negatively with MMSE; E. coli positively correlated with curli and β-amyloid; HDL-C negatively correlated. LPS inversely correlated with F. prausnitzii.

Findings support the hypothesis that age-related dysbiosis contributes to cognitive decline through systemic inflammation and microbial products that may influence amyloid and tau pathology. Elevated serum LPS with aging/MCI/dementia and its alignment with predicted LPS biosynthesis in the microbiome suggest compromised gut barrier and immune activation along the gut–brain axis. Increased E. coli and its amyloid curli in stool, together with correlations to lower MMSE and higher β-amyloid/tau, implicate bacterial amyloids as potential facilitators of host amyloid aggregation. The rat model, absent metabolic comorbidities, showed aging-associated increases in E. coli, fecal and brain curli, LPS, TNFα, and oxidative stress markers, supporting a causal link between aging microbiome changes and neuroinflammation/amyloidogenic environment. The inverse association of Faecalibacterium prausnitzii with LPS underscores the potential protective role of anti-inflammatory commensals. Collectively, the data indicate that microbiota composition and function, particularly Gram-negative taxa and bacterial amyloids, may modulate neurodegeneration risk, and that circulating LPS, β-amyloid, tau, BCAA, and fecal curli could serve as accessible biomarkers related to cognitive impairment.

Dysbiosis of the gut microbiota in aging is associated with increased serum LPS, β-amyloid, and tau, and elevated fecal curli protein, with higher E. coli abundance. In aging rats, curli protein also increased in brain alongside markers of neuroinflammation and oxidative stress. These human and animal findings suggest a gut–brain axis mechanism wherein bacterial products (LPS and curli) may promote amyloid/tau pathology and cognitive decline. The identified markers may aid early detection of MCI and dementia, but further studies in larger cohorts are needed to validate their utility and to clarify mechanisms.

Human data are cross-sectional, limiting causal inference. The dementia group sample size was relatively small (n = 13). Curli protein could not be measured in human brain; brain analyses were performed only in rats. Participants were from a specific geographic region, and generalizability requires larger, diverse cohorts. Functional pathway inferences (PICRUSt) are predictive rather than direct metagenomic/metatranscriptomic measurements. Comorbidities (e.g., diabetes) may confound biomarker levels.