Live and Heat-Killed Probiotic Protects the Aging Mouse Brain and Blood-Brain Barrier
A kimchi-derived probiotic, live or heat-killed, cut AGEs, brain inflammation and blood-brain barrier leakage in accelerated-aging mice and shifted gut bacteria.
Summary
Researchers sped up aging in mice with daily D-galactose injections for 8 weeks. This raised blood levels of damaging sugar-protein compounds called AGEs, made the blood-brain barrier leaky, and inflamed the hippocampus, the brain's memory center. Mice also given a kimchi-derived probiotic, Lactiplantibacillus plantarum HY7715, fared much better. Whether the bacteria were alive or heat-killed, AGE levels fell by about 60%, dye leakage into the brain dropped by roughly 80%, and inflammation, cell-senescence and cell-death signals eased. The probiotic also strengthened the gut lining and raised Bifidobacterium in the gut. These results suggest a gut-to-brain link in aging. They come from mice only, with no memory or behavior testing, so they are early evidence rather than advice for people.
Detailed Summary
The blood-brain barrier (BBB) tightly controls what reaches brain tissue, and it weakens with age as oxidative stress, inflammation and glycation damage its tight junction proteins. Gut changes with aging may feed this process through systemic inflammation. Whether a probiotic can interrupt the chain from gut microbes to AGEs to BBB leakage to neuroinflammation has been unclear, which motivated this study of the kimchi-derived strain Lactiplantibacillus plantarum HY7715.
Male C57BL/6J mice (7 weeks old, 10 per group) were split into four groups: normal control, D-galactose control (150 mg/kg/day subcutaneously for 8 weeks), D-galactose plus live HY7715, and D-galactose plus heat-killed HY7715. Both bacterial preparations were dosed orally at the equivalent of 1 × 10^9 CFU/kg/day. Seven mice per group were used for biochemical, molecular and imaging work, and three per group for the Evans blue leakage assay. Researchers also measured gene expression in the cortex, hippocampus and colon, and sequenced fecal 16S rRNA.
D-galactose raised serum AGEs from 5.26 to 17.33 μg/mL. Live and heat-killed HY7715 brought them down to 6.71 and 6.12 μg/mL (p<0.001). Brain Evans blue leakage rose from 0.05 to 0.39 OD/g with D-galactose, then fell to 0.08 and 0.07 OD/g with treatment (p<0.001). Cortical claudin-5 staining fell from 1.75% to 1.03% of area and recovered to 1.70% (live) and 2.06% (heat-killed). Cldn5, Tjp1 and Ocln mRNA were also restored.
In the hippocampal CA1 region, Iba1-positive microglia rose from 105 to 149 cells/mm² and returned to about 110 and 105 with treatment. Hippocampal IL-1β mRNA climbed 53-fold and IL-6 23-fold with D-galactose. Treated groups showed IL-1β at 3.25- and 2.62-fold and IL-6 at 2.55- and 3.09-fold (p<0.001). Senescence markers p53 and p21 fell (p21 from 2.17- to 1.31- and 1.21-fold). Antioxidant genes (Gpx1, Gpx2, Sod1, Sod2) were restored, and Bax dropped from 1.85- to about 1.15-fold while Bcl-2 rose.
In the colon, tight junction genes Tjp1 and Ocln recovered, while the permeability-linked Cldn2 fell from 1.66- to about 1.24-fold. Colonic TNF-α, IL-1β and IL-6 also decreased (p<0.001). Alpha diversity did not differ between groups. Live HY7715 significantly shifted community structure versus D-galactose controls (unweighted UniFrac p=0.023; weighted p=0.010), whereas heat-killed did not. Both treatments increased Bifidobacterium. Live HY7715 lowered Desulfovibrio and raised Bifidobacterium pseudolongum. Bifidobacterium correlated negatively with serum AGEs, while Evans blue leakage correlated positively with Desulfovibrio and negatively with Akkermansia.
The results suggest that probiotics, even non-viable ones, may act on the gut-AGE-BBB-brain axis. That heat-killed bacteria worked as well as live ones points to bacterial components rather than colonization. Caveats are substantial: this is a chemically induced aging model in male mice only, with no cognitive testing, mostly mRNA readouts, and correlational microbiome data. It does not show that gut changes cause the brain benefits.
Key Findings
- Serum AGEs rose from 5.26 to 17.33 μg/mL with D-galactose, then fell to 6.71 (live) and 6.12 μg/mL (heat-killed) with HY7715 (p<0.001)
- Brain Evans blue leakage dropped from 0.39 to 0.08 (live) and 0.07 OD/g (heat-killed), about an 80% reduction in BBB permeability (p<0.001)
- Cortical claudin-5-positive area fell from 1.75% to 1.03% with D-galactose and rose to 1.70% (live) and 2.06% (heat-killed); Cldn5, Tjp1 and Ocln mRNA were also restored
- Hippocampal IL-1β mRNA fell from 53.2-fold to 3.25- (live) and 2.62-fold (heat-killed) vs normal controls; Iba1-positive microglia dropped from 149 to about 110 and 105 cells/mm²
- Senescence and apoptosis markers improved: p21 fell from 2.17- to 1.31- and 1.21-fold and Bax from 1.85- to about 1.15-fold, while Gpx1/2 and Sod1/2 were restored
- In the colon, Cldn2 fell from 1.66- to about 1.24-fold and TNF-α, IL-1β and IL-6 all decreased (p<0.001), alongside recovery of Tjp1 and Ocln
- Both treatments raised Bifidobacterium; only live HY7715 significantly shifted overall microbiota structure (weighted UniFrac p=0.010), and Bifidobacterium correlated inversely with serum AGEs
Methodology
Male C57BL/6J mice (7 weeks old; 10 per group, four groups) received subcutaneous D-galactose (150 mg/kg/day) for 8 weeks, with oral live or heat-killed L. plantarum HY7715 at the equivalent of 1 × 10^9 CFU/kg/day. Seven mice per group were used for ELISA, RT-qPCR and immunofluorescence, and three per group for Evans blue extravasation. Fecal microbiota were profiled by 16S V3-V4 sequencing (QIIME 2, LEfSe, UniFrac/PERMANOVA) and linked to phenotypes by Spearman correlation. Groups were compared with one-way ANOVA and Dunnett's test, with p<0.05 considered significant.
Study Limitations
The D-galactose model reproduces only some features of natural aging, and the study used male mice only, one dose and no behavioral or cognitive testing. Evans blue assays used just three mice per group, imaging used one section per animal, and most molecular readouts were mRNA rather than protein. Microbiome results are correlational, dominated by Faecalibaculum (>70%), and live HY7715 only was associated with significant community shifts; species-level calls were exploratory. The strain was produced by hy Co., Ltd., and the study was approved by that company's animal committee, which suggests industry involvement; no independent replication was reported. The supplied full text was also cut off in the Discussion, so any further limitations the authors stated could not be reviewed.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: