Alzheimer's disease (AD) is a progressive neurodegenerative disorder with limited therapeutic options. Emerging evidence highlights the gut-brain axis as a promising target, and natural polysaccharides offer multi-target advantages. Here, we isolated and structurally characterized PF30-3, a homogeneous branched α-D-glucan (69.91 kDa) from Pseudostellaria heterophylla, featuring a (1 → 4)-α-d-glucopyranosyl backbone with (1 → 4,6)-α-d-glucopyranosyl branching points. In LPS-stimulated RAW264.7 macrophages, 200 μg/mL PF30-3 significantly inhibited nitric oxide release, restored cell viability, and rebalanced inflammatory cytokines. The in vivo efficacy was first validated in an Aβ₁₋₄₂-induced zebrafish AD-like model, where 20 mg/kg PF30-3 effectively rescued spatial memory and learning impairments. These findings were further translated to 5 × FAD mice, in which oral administration of 50 mg/kg PF30-3 for 24 consecutive days significantly improved the recognition and spatial memory without observable acute toxicity. Mechanistically, PF30-3 reduced cerebral Aβ deposition in vivo, an effect not attributable to direct aggregation inhibition as shown by ThT and CD assays. Instead, PF30-3 exerted its effects through multiple indirect pathways, including inhibition of glial cell hyperactivation, repair of intestinal barrier integrity, rebalancing of serum inflammatory cytokines, and remodeling of gut microbiota composition, characterized by a decreased Firmicutes/Bacteroidota ratio and enrichment of beneficial genera such as Muribaculum, Bacteroides, and Prevotellaceae_UCG-001. Correlation analyses linked these microbial shifts to improved cognitive outcomes and reduced neuroinflammation. Collectively, these findings demonstrate that PF30-3 exerts neuroprotective effects through modulation of the microbiota-gut-brain axis, highlighting it as a promising natural polysaccharide-based candidate for AD therapy.
