Bacteria-derived glutarate mitigates Alzheimer’s disease model pathology through autophagy-lysosomal pathway Authors: Cui Deng, Mingqin Qu, Wei Yao, Xiaotong Hu, Yan Li, Weilun Zheng, Yiying Zhai, Fangyan Liu, Ting Yang, Tianlong Wang, Xiangming Wang Published: 2026 ABSTRACT Probiotics exert neuroprotective effects against Alzheimer’s disease (AD) by modulating gut-brain axis pathways, though disease-modifying therapies remain unavailable. Our study revealed that Escherichia coli (E. coli) strain HB101 ameliorated AD-related phenotypes in Caenorhabditis elegans‌ (C. elegans) models, including learning deficits, neurodegeneration, and paralysis. Mechanistically, HB101 reduced amyloid-β (Aβ) aggregation by enhancing lysosomal activity, autophagy, and mitochondrial/endoplasmic reticulum unfolded protein responses (UPRmt/UPRer). Specifically, HB101 activated UPRmt via atfs-1 and sphk-1, and UPRer through pek-1. Metabolomic screening pinpointed glutarate as a bioactive metabolite that mitigates AD-related pathology through lysosomal activation and autophagy promotion.Graphical Abstract Introduction Alzheimer’s disease (AD), the most prevalent form of neurodegenerative dementia, is characterized by complex pathological mechanisms including amyloid-β (Aβ) plaque accumulation, neurofibrillary tangles of hyperphosphorylated tau protein, and chronic neuroinflammation [1, 2]. Currently, there are no disease-modifying therapies approved for clinical use. Identifying reliable biomarkers for early diagnosis and precise disease staging is essential for advancing AD research and improving clinical management [3, 4]. Despite the development of symptom-alleviating treatments such as donepezil and memantine, as well as emerging approaches like terahertz technology [5], current therapeutic options for AD remain limited to symptom management rather than disease modification. The urgent need persists for novel therapeutic strategies capable of halting or reversing disease progression to achieve meaningful cure for AD. Recent years have witnessed the growing prominence of the “microbiota-gut-brain axis” (MGBA) theory, which has underscored the pivotal role of gut microbiota homeostasis [6, 7] in neurological health. This paradigm shift has catalyzed substantial academic enthusiasm for investigating probiotics as a disease-modifying strategy against AD. Given their well-documented pleiotropic benefits, probiotics emerge as a particularly promising therapeutic avenue for targeting AD pathogenesis. Emerging evidence suggests that probiotics may exert neuroprotective effects through modulation of the gut-brain axis, potentially via anti-inflammatory pathways, regulation of Aβ metabolism, promotion of neurotrophic factor expression, and enhancement of blood-brain barrier integrity [8]. These mechanisms highlight probiotics as a promising therapeutic avenue for AD intervention. Emerging evidence suggests that diverse probiotics, including Lactobacillus spp., Bifidobacterium spp., and Bacillus subtilis, exhibit efficacy in mitigating AD phenotypes across murine and C. elegans models [9], mechanistic insights remain elusive. Similarly, Lacticaseibacillus rhamnosus, Pediococcus parvulus, Bacillus Subtilis, and Lactobacillus paracasei have shown AD-modulatory effects in C. elegans models [10–12], though their precise molecular targets require further investigation. Short-chain fatty acids, indole derivatives and bile acids have been identified as functional metabolites of probiotics that ameliorate AD phenotypes [13]. Anti-inflammation and anti-oxidation are well-established common mechanisms by which these metabolites alleviate AD progression. Nevertheless, novel bacterial-derived functional metabolites with anti-AD effects remain to be further explored, and their underlying molecular mechanisms require in-depth investigation. Here, we demonstrate that E. coli HB101 ameliorates AD model phenotypes in C. elegans, including learning deficits, neurodegeneration, and paralysis, by mitigating Aβ aggregation through the enhancement of lysosomal function, autophagy, and the activation of UPRmt (via atfs-1 and sphk-1) and UPRer (via pek-1). Metabolomic profiling identified glutarate as a potential metabolite that promotes lysosomal function and autophagy to counteract AD-related pathology. Materials and methods C. elegans strains and culture All C. elegans strains were cultured at 20 °C, with the exception of the CL2331 strain, which was maintained at 15 °C. All worms were cultivated on standard NGM (Nematode Growth Medium) agar plates [14]. The specific C. elegans strains employed in this study are listed below. strainsgenotypeN2Wild typeCL2006dvIs2 [Punc-54::A-beta 1-42; rol-6 (su1006)]CL2331dvIs37 [Pmyo-3::GFP::A-beta 3–42; rol-6(su1006)]CX3553kyIs104 [Pstr-1::GFP]GMC101dvIs100 [Punc-54::A-beta 1-42; Pmtl-2::GFP]GRU101gnaIs1 [Pmyo-2::YFP]GRU102gnaIs2 [Pmyo-2::YFP+Punc-119::A-beta 1-42]HZ3431HLH-30::GFPMAH215sqIs11 [Plgg-1::mCherry:: GFP::LGG-1; rol-6]OH15876pha-4 (ot946 [pha-4::3xGAS::GFP::TEV::LoxP::3xFLAG])SJ4100zcIs13 [Phsp-6::GFP; lin-15(+)]SJ4005zcIs4 [Phsp-4::GFP]LTY408SJ4100; atfs-1(gk3094)TV53181[Punc-54::GFP]TV53221dvIs100; pqm-1 (ok485)TV53239Punc-54::A-beta 1–42; Pord-1::RFP; juIs76TV53244GRU102; cmuIs2[Peat-4::iGluSnFR]TV53067Punc-30::NAD+ sensorTYL001Pcpr-5::GFP; rol-6XMW184dvIs100; pek-1 (ok275)XMW188dvIs100; sphk-1 (ok1097)XW19180Phsp-16.2::NUC-1::pHTomato bpIs151 [Psqst-1::SQST-1::GFP; unc-76(+)] cmuIs1 [Peat-4::iGluSnFR; Pmyo-2::mCherry]XMW274TYL001; atfs-1(gk3094)XMW278MAH215; atfs-1(gk3094) strainsgenotypeN2Wild typeCL2006dvIs2 [Punc-54::A-beta 1-42; rol-6 (su1006)]CL2331dvIs37 [Pmyo-3::GFP::A-beta 3–42; rol-6(su1006)]CX3553kyIs104 [Pstr-1::GFP]GMC101dvIs100 [Punc-54::A-beta 1-42; Pmtl-2::GFP]GRU101gnaIs1 [Pmyo-2::YFP]GRU102gnaIs2 [Pmyo-2::YFP+Punc-119::A-beta 1-42]HZ3431HLH-30::GFPMAH215sqIs11 [Plgg-1::mCherry:: GFP::LGG-1; rol-6]OH15876pha-4 (ot946 [pha-4::3xGAS::GFP::TEV::LoxP::3xFLAG])SJ4100zcIs13 [Phsp-6::GFP; lin-15(+)]SJ4005zcIs4 [Phsp-4::GFP]LTY408SJ4100; atfs-1(gk3094)TV53181[Punc-54::GFP]TV53221dvIs100; pqm-1 (ok485)TV53239Punc-54::A-beta 1–42; Pord-1::RFP; juIs76TV53244GRU102; cmuIs2[Peat-4::iGluSnFR]TV53067Punc-30::NAD+ sensorTYL001Pcpr-5::GFP; rol-6XMW184dvIs100; pek-1 (ok275)XMW188dvIs100; sphk-1 (ok1097)XW19180Phsp-16.2::NUC-1::pHTomato bpIs151 [Psqst-1::SQST-1::GFP; unc-76(+)] cmuIs1 [Peat-4::iGluSnFR; Pmyo-2::mCherry]XMW274TYL001; atfs-1(gk3094)XMW278MAH215; atfs-1(gk3094) Bacterial strains and growth conditions The E. coli strains utilized in this study were OP50 and HB101. Initially, the glycerol-stocked bacterial cultures were retrieved from the − 80 °C freezer and aseptically transferred to a biosafety cabinet. A sterile pipette tip was used to inoculate the frozen glycerol stock, which was then streaked in a zigzag pattern onto sterile NGM agar plates. Following streaking, the plates were incubated at 37 °C for 24 h. Subsequently, a single well-isolated colony was selected under sterile conditions and inoculated into 40 mL of LB broth. The liquid culture was incubated at 37 °C for an additional 24 h with shaking. Following this incubation period, the resulting bacterial suspension was spread-plated onto fresh NGM agar plates using a sterile glass spreader. The prepared plates were then kept at room temperature for 24 h prior to experimental use. Paralysis assay Paralysis assays were performed using C. elegans strains GMC101 and CL2006, which were cultured at 20 °C and 15 °C respectively. Synchronized L4-stage worms (n = 4) were initially transferred to NGM plates seeded with either OP50 or HB101 E. coli and allowed to grow for four days. Subsequently, approximately 30 healthy L4-stage worms were selected and transferred to fresh NGM plates containing the same bacterial strain. The plates were then subjected to a 24-hour heat shock treatment at 25 °C in a temperature-controlled incubator. Paralysis was assessed 24 h post-treatment using a standardized scoring criterion: worms were considered paralyzed if they failed to respond to gentle head stimulation with a platinum wire (no movement or failure to move away within 3 s) or exhibited only slight head movements without body displacement upon stimulation. The paralysis rate was calculated as the percentage of paralyzed worms relative to the total number of worms tested. Non-associative learning assay First, four L4-stage C. elegans were picked and placed on NGM plates seeded with OP50 or HB101 E. coli for growth. After four days, a large number of L4-stage progeny worms would be produced; approximately 30 healthy L4-stage worms were then selected and transferred to new NGM plates containing OP50 or HB101. One day later, when the worms developed into the Day 1 adult stage, the learning and memory assay was initiated. First, the worms were picked and transferred to blank NGM plates for a 30-second acclimation period. Subsequently, the head of each worm was gently touched with an eyebrow hair; upon this stimulus, the worm would move backward. When the worm finished retreating and began to resume forward movement, gentle touch with the eyebrow hair was repeated until the worm no longer exhibited a retreat response. The total number of retreats made by each worm was recorded. All of these experiments were conducted at least three times. Food choice assay A bacterial preference assay was conducted using a 3 cm-diameter NGM agar plate. A diameter line was drawn through the center of the plate to serve as a reference. Symmetrically along this diameter line, 10 µL aliquots of OP50 and HB101 bacterial suspensions (adjusted to equal colony-forming units, CFUs) were spotted at positions 1/4 of the plate radius from the center. Fifty synchronized L4-stage GMC101 C. elegans were then carefully transferred to the center of the plate. The number of worms that migrated toward the OP50 and HB101 bacterial spots, respectively, was recorded after a defined incubation period. Fecundity assay To assess reproductive output, individual L4-stage GMC101 C. elegans (n = 3 per condition) were transferred to NGM plates seeded with either OP50 or HB101 E. coli as a food source. Egg-laying was monitored daily beginning on Day 1 of adulthood (24 h post-L4 stage). Each day, after recording the number of eggs laid, the worms were transferred to fresh NGM plates with the corresponding bacterial strain to standardize food availability. This daily transfer and egg-counting procedure were continued until the worms ceased egg production, allowing for the calculation of the total lifetime fecundity for each individual. All of these experiments were conducted three times. Growth rate assay To assess developmental progression, 50 µL of alkaline hypochlorite solution (0.5% sodium hypochlorite, 0.5 mol/L NaOH) was applied to food-free regions of NGM plates seeded with either OP50 or HB101 E. coli. Five Day 1 adult worms were then introduced into each solution. After a 3-day incubation period at 20 °C, the developmental stage of the worms was assessed, and the number of individuals that had progressed to the L4 stage or later was recorded separately from those that remained in earlier developmental stages. The proportion of worms reaching the L4 stage was then calculated and expressed as a percentage. All of these experiments were conducted at least three times. Motor ability assay To evaluate locomotor capacity, four L4-stage worms were initially transferred to NGM plates seeded with either OP50 or HB101 E. coli for growth. After four days of cultivation, approximately 25 healthy L4-stage progenies were transferred to fresh NGM plates containing the respective bacterial strain. Upon reaching Day 1 adulthood, locomotor activity was assessed. For each assay, a single worm was placed in a 200 µL M9 buffer droplet on a glass slide and allowed to acclimate for 30 s. Locomotor activity was quantified by counting the number of complete body waves (defined as one “oscillation”) executed within a 30-second observation period. Partial movements involving only the head or tail were excluded from the count to ensure consistency in measuring full-body motility. Heat stress assay To assess heat stress resistance, L4-stage worms were initially cultured on NGM plates seeded with either OP50 or HB101 E. coli for four days. Subsequently, at least 30 L4-stage progenies from each group were transferred to fresh plates and cultured for an additional 24 h. For the N2 wild-type strain, worms were subjected to a 4-hour heat shock at 35 °C, while the GMC101 strain was exposed to 30 °C for 24 h. Following the respective heat treatments, survival rates were quantified and statistically analyzed to evaluate strain-specific thermotolerance. Oxidative stress resistance assay To assess oxidative stress resistance, synchronized L4-stage worms were first cultured for four days on NGM agar plates pre-seeded with either OP50 or HB101 E. coli strains. Following this initial growth period, approximately 30 healthy L4-stage worms from each group were transferred to fresh NGM plates containing the corresponding bacterial strain. After an additional 24-hour incubation period at 20 °C, the worms were washed two to three times with M9 buffer to eliminate residual bacteria from their cuticles. The washed worms were then exposed to oxidative stress by transferring them to M9 buffer supplemented with 50 mM hydrogen peroxide (H2O2) and incubated in the dark at 20 °C for 30 min. Post-treatment survival was determined by gentle mechanical stimulation with a platinum wire, with viability confirmed by observing movement responses. The survival rate was calculated as the percentage of live worms exhibiting a positive movement response relative to the total number of worms tested. Lifespan assay To conduct a comprehensive lifespan analysis, GMC101 worms were synchronized to obtain a population of worms at the same developmental stage. Following synchronization, ≥ 50 L4-stage worms were inoculated onto NGM plates pre-seeded with either OP50 or HB101 E. coli for standard culturing. The experiment was performed under controlled conditions (20 °C) with daily monitoring. At 24-hour intervals, the number of deceased worms was recorded, and surviving individuals were transferred to fresh NGM plates to prevent overcrowding and ensure continuous food availability. This process was repeated until all worms in the population had reached the end of their lifespan [15]. The entire assay was conducted twice to ensure statistical reliability, and survival curves were generated using Kaplan-Meier analysis. The lifespan experiment was repeated at least twice. HB101 functional part assay To investigate the bioactive components of the bacteria, single colonies of OP50 or HB101 were inoculated into 40 mL of LB broth and cultured with shaking at 37 °C for 24 h. The bacterial suspension was centrifuged at 13,000 rpm for 5 min to collect the bacterial pellets, and the supernatant was further centrifuged to remove residual bacteria before sterilization through 0.22 μm membrane filtration. For the OP50 pellet group, the pellet was resuspended in 500 µL of ultrapure water, and 10 µL of this suspension was mixed with the sterilized HB101 supernatant, followed by spreading 60 µL of the mixture onto NGM plates. For the heat-inactivated HB101 pellet group, the pellet was autoclaved at 121 °C for 20 min or heat-treated at 70 °C for 30 min, resuspended in 500 µL of ultrapure water, and 10 µL of this suspension was mixed with the OP50 suspension before plating 60 µL onto NGM plates. All procedures were replicated at least three times. Fluorescence microscopy and visualization assay ‌A 3% (w/v) agarose solution was prepared by microwave heating until fully dissolved, followed by brief cooling to ~ 50 °C. A small aliquot of molten agarose was dispensed onto a glass slide using a weighing spoon, and a second glass slide was used to press the solution into a uniform pad (~ 1 mm thickness). After 10 s of solidification, the upper slide was carefully removed to expose the agarose pad.‌ A 4× working solution of levamisole was prepared by diluting a 10× stock solution (1 mg/mL) in ultrapure water. A 2.5 µL aliquot of this solution was pipetted onto the center of the agarose pad. Worms at the appropriate developmental stage were picked using a platinum wire pick, transferred to the anesthetic-treated agarose pad, and fully immersed. A coverslip was then gently placed over the sample to ensure even distribution. Fluorescence imaging was performed using a Zeiss Imager M2 microscope with filter sets 20 (Rhodamine) and 38 (Endow GFP). Acquired images were analyzed using ImageJ software to quantify fluorescence intensity, fluorescence imaging was performed using a Zeiss LSM 980 confocal microscope with Airyscan 2. Metabolome analysis Single colonies of OP50 or HB101 E. coli were inoculated into 40 mL of LB medium and incubated at 37 °C for 24 h. The cultured bacterial suspension was centrifuged at 13,000 rpm for 5 min, the supernatant was discarded, and the centrifugation was repeated under the same conditions. The resulting pellet was resuspended and washed with ultrapure water, followed by another centrifugation step. This washing process was repeated three times. Finally, the bacterial pellet was stored at -80 °C. The entire procedure was independently repeated, yielding a total of 3 samples, which were sent to Shanghai Shenggong Biological Engineering Co., Ltd. for metabolomic analysis. The instruments used for metabolomic analysis were an ultra-high-performance liquid chromatograph (Model: LC-30 A, Brand: Shimadzu, Origin: Japan) and a mass spectrometer (Model: TripleTOF 6600+, Brand: SCIEX, Origin: Foster City, California, USA). All samples were analyzed via two LC/MS detection methods: one portion of the sample was analyzed in positive ion mode, paired with a T3 chromatographic column (Waters ACQUITY Premier HSS T3 column, specifications: 1.8 μm, 2.1 mm × 100 mm); mobile phase A was an aqueous solution containing 0.1% formic acid, and mobile phase B was an acetonitrile solution containing 0.1% formic acid. The elution gradient was set as follows: the proportion of mobile phase B was increased from 5% to 20% within the first 2 min, then raised to 60% over the next 3 min, subsequently increased to 99% within 1 min and maintained for 1.5 min, and finally decreased to 5% within 0.1 min and held for 2 min. The analytical conditions were: column temperature of 40 °C, mobile phase flow rate of 0.4 mL/min, and injection volume of 4 µL. The other portion of the sample was analyzed in negative ion mode, with the elution gradient identical to that used in the positive ion mode. Drug treatment Glutarate and MG-132 were dissolved in ultrapure water to prepare a 50 mM stock solution, Rapamycin is 100 mM stock solution, which was stored at -20 °C for later use. Prior to use, the stock solution was diluted to a working concentration of 10 µM using OP50 E. coli liquid culture. For plate preparation, 60 µL of the diluted mixture was spread onto NGM plates. Statistical analysis Statistical analyses were performed using GraphPad Prism8.0.2 software. For comparisons between two groups, two-tailed unpaired t-tests were used to assess the significance of differences; for multi-group analyses, one-way analysis of variance (one-way ANOVA) was employed to detect significant differences. Tukey’s multiple comparisons test was applied to correct for multiple comparisons following one-way ANOVA. Categorical data were analyzed using Fisher’s exact test [16]. All analyses were conducted with a conventional significance level of α = 0.05 (P < 0.05 was considered statistically significant). Statistically significant results in all data were marked using standardized graphical notations: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, while “ns” (not significant) indicated no statistical significance (P > 0.05). Results The E. coli HB101 ameliorates the neurodegeneration of the Aβ AD model The expression of Aβ1–42 in body wall muscle cells of C. elegans generates the transgenic strain GMC101 (hereafter referred to as Aβ), which promotes oligomer and aggregate formation. These pathological alterations induce severe age-progressive paralysis, manifesting predictably with organism aging [17]. Using this typical phenotype, we found that E. coli HB101 ameliorated paralysis in the C. elegans Aβ AD model (Fig. 1A). To validate this protective effect, we employed an independent Aβ AD model co-expressing Aβ1–42 in body wall muscle cells with a roller phenotype marker (CL2006). Consistent with our initial findings, HB101 treatment demonstrated comparable paralysis alleviation in this alternative model (Fig. 1A), thereby confirming its ability to counteract Aβ toxicity. Since the unc-54 promoter drove Aβ expression in these AD models, we assessed whether HB101 might influence promoter activity using a Punc-54::GFP reporter strain. No significant difference in GFP fluorescence was observed between worms fed HB101 and those fed the control strain OP50 (Fig. 1B, C), suggesting that HB101 mitigates AD-related phenotypes by specifically targeting Aβ rather than modulating unc-54 promoter activity. Fig. 1The E. coli HB101 ameliorates the neurodegeneration of the Aβ AD model. A The scatter plot shows the paralysis rate of the two Aβ AD models treated with the control E. coli OP50 or HB101. Two-tailed unpaired Student’s t-test. n = 3. B-C Representative fluorescence images and quantitative analysis of fluorescence intensity of Punc-54::GFP treated with OP50 or HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 20. D Scatter plot showing the number of retreats of the AD model GRU102 and the control GRU101 (left). Scatter plot showing the number of retreats of GRU102 treated with OP50 or HB101 (right). Two-tailed unpaired Student’s t-test. n = 24-26. E-F Representative fluorescence images and quantitative analysis of commissures in the Aβ AD model and the control. Arrows indicate commissures. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 26-33. G-H Representative fluorescence images and quantitative analysis of commissures in the Aβ AD model treated with OP50 or HB101. Arrows indicate commissures. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 23-24. I Scatter plot showing the total number of eggs laid of the Aβ AD model treated with OP50 or HB101. Two-tailed unpaired Student’s t-test. n = 3. J Scatter plot showing the growth rate of the Aβ AD model treated with OP50 or HB101. Two-tailed unpaired Student’s t-test, n = 3. K Scatter plot showing food preference of the Aβ AD model. Two-tailed unpaired Student’s t-test, n = 3. L Scatter plot illustrating the impact of HB101’s pasteurization, supernatant and autoclaving on the Aβ AD model. One-way ANOVA followed by Tukey’s post hoc test, n = 3. M Scatter plot showing heat resistance of HB101 in the Aβ AD model or N2. Two-tailed unpaired Student’s t-test, n = 3-5. N Scatter plot showing the thrashing per 30s of HB101 in the Aβ AD model or N2. Two-tailed unpaired Student’s t-test. n = 22-25. O The survival curve of the Aβ AD model treated with OP50 or HB101. Scatter plot showing the mean lifespan of the Aβ AD model. Log-rank test for lifespan, Two-tailed unpaired Student’s t-test. n = 50-62. Data are presented as mean ± SEM. ns, not significant (P > 0.05); *P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001 The E. coli HB101 ameliorates the neurodegeneration of the Aβ AD model. A The scatter plot shows the paralysis rate of the two Aβ AD models treated with the control E. coli OP50 or HB101. Two-tailed unpaired Student’s t-test. n = 3. B-C Representative fluorescence images and quantitative analysis of fluorescence intensity of Punc-54::GFP treated with OP50 or HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 20. D Scatter plot showing the number of retreats of the AD model GRU102 and the control GRU101 (left). Scatter plot showing the number of retreats of GRU102 treated with OP50 or HB101 (right). Two-tailed unpaired Student’s t-test. n = 24-26. E-F Representative fluorescence images and quantitative analysis of commissures in the Aβ AD model and the control. Arrows indicate commissures. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 26-33. G-H Representative fluorescence images and quantitative analysis of commissures in the Aβ AD model treated with OP50 or HB101. Arrows indicate commissures. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 23-24. I Scatter plot showing the total number of eggs laid of the Aβ AD model treated with OP50 or HB101. Two-tailed unpaired Student’s t-test. n = 3. J Scatter plot showing the growth rate of the Aβ AD model treated with OP50 or HB101. Two-tailed unpaired Student’s t-test, n = 3. K Scatter plot showing food preference of the Aβ AD model. Two-tailed unpaired Student’s t-test, n = 3. L Scatter plot illustrating the impact of HB101’s pasteurization, supernatant and autoclaving on the Aβ AD model. One-way ANOVA followed by Tukey’s post hoc test, n = 3. M Scatter plot showing heat resistance of HB101 in the Aβ AD model or N2. Two-tailed unpaired Student’s t-test, n = 3-5. N Scatter plot showing the thrashing per 30s of HB101 in the Aβ AD model or N2. Two-tailed unpaired Student’s t-test. n = 22-25. O The survival curve of the Aβ AD model treated with OP50 or HB101. Scatter plot showing the mean lifespan of the Aβ AD model. Log-rank test for lifespan, Two-tailed unpaired Student’s t-test. n = 50-62. Data are presented as mean ± SEM. ns, not significant (P > 0.05); *P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001 Impaired learning and memory are cardinal features of AD. To assess the ameliorative effects of HB101 on these deficits, we employed a well-established non-associative learning paradigm in C. elegans, habituation behavior, where responses to repeated stimuli progressively diminish over time [18]. Using the pan-neuronal Aβ expression AD model (GRU102), we observed that it required significantly more touches to habituate compared to the control GRU101 worms (Fig. 1D), indicative of learning deficits. Importantly, HB101 treatment significantly reduced the number of touches required for habituation in GRU102 worms (Fig. 1D), demonstrating its efficacy in alleviating AD-related learning impairments. Given that neurodegeneration represents a core pathological feature of AD, we sought to determine whether HB101 could attenuate neuronal loss in AD model. To circumvent potential confounding effects from muscle-expressed GFP (as observed in GMC101), we crossed an independent Aβ-expressing line with Punc-25::GFP, a specific marker for D-type motor neurons [19]. Quantitative analysis revealed that the Aβ-expressing worms exhibited significantly fewer commissures, the axonal connections between ventral and dorsal nerve cords, compared to control animals (Fig. 1E-F). Notably, HB101 administration restored commissure numbers to near-physiological levels (Fig. 1G-H), providing compelling evidence for its neuroprotective efficacy against Aβ-induced neurodegeneration. Next, we sought to assess whether HB101 affects the overall physiology of C. elegans. First, treating the Aβ AD model with HB101 did not change the number of offspring produced (Fig. 1I), indicating that worm fertility remains unaffected. Second, HB101 treatment did not alter the time taken for development from egg to the larval stage 4 (L4) (Fig. 1J), suggesting that the growth rate in the Aβ AD model is unchanged. Third, in a food preference test, no significant difference was found between HB101 and the commonly used food source OP50 (Fig. 1K), showing that worms do not prefer one over the other. To determine if HB101’s effect depends on live bacteria, we fed the Aβ AD model with HB101 that had been killed by autoclaving or pasteurization. The results indicated that the dead HB101 lost its protective effect (Fig. 1L), suggesting that live bacteria are required for protection or that the active compounds are not heat-stable. We then examined whether the protective effect could be reproduced without bacteria by adding HB101 culture supernatant to the worms’ OP50 diet. The findings showed that the HB101 supernatant did not decrease paralysis in the Aβ AD model (Fig. 1L), implying that the active compounds may not be secreted. We further examined if HB101 could improve stress resistance in C. elegans. HB101 significantly increased survival under heat stress in wild-type N2 and the Aβ AD model worms compared to OP50 controls (Fig. 1M). Lastly, we evaluated whether HB101 had effects beyond its neuroprotective role. HB101 treatment did not significantly change thrashing behavior in M9 buffer for either the Aβ AD model or wild-type N2 worms compared to OP50 controls (Fig. 1N), but it notably extended lifespan in the Aβ AD model (Fig. 1O). HB101 mitigates Aβ aggregation by simultaneously enhancing lysosomal function, autophagy, UPRmt and UPRer Aβ plaques represent a pathological hallmark of AD, making the development of therapeutic strategies aimed at reducing Aβ aggregation a major research focus. To quantify Aβ aggregation, we utilized a transgenic C. elegans strain (CL2331) that expressed muscle-specific GFP-tagged Aβ (GFP::Aβ), which formed distinct aggregation foci (Fig. 2A). Notably, treatment with HB101 significantly reduced these aggregates compared to OP50-fed controls (Fig. 2A), demonstrating its capacity to mitigate AD pathology by suppressing Aβ aggregation. To elucidate the underlying mechanism, we subsequently investigated the molecular pathway mediating HB101’s anti-aggregation effects. Fig. 2HB101 mitigates Aβ aggregation by simultaneously enhancing lysosomal function, autophagy, UPRmt and UPRer. A Representative fluorescence images and quantitative analysis of the number of Aβ aggregation plaques treated with OP50 or HB101. Arrows indicate Aβ aggregation plaques. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 11–15. B Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP treated with HB101 or not. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 23–24. C Representative fluorescence images and quantitative analysis of fluorescence intensity of single lysosome treated with OP50 or HB101. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 51–135. D Representative fluorescence images and quantitative analysis of the number of autophagic substrates SQST-1::GFP treated with OP50 or HB101. Arrows indicate autophagic substrates. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 21–25. E Representative fluorescence images and quantitative analysis of the number of HLH-30::GFP positive nuclei treated with OP50 or HB101. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 30. F Representative fluorescence images and quantitative analysis of the total number of PHA-4::GFP positive nuclei at the pharyngeal bulb. Arrows indicate nucleus. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 21–23. G Representative fluorescence images and quantitative analysis of the red only puncta in the head of mCherry::GFP::LGG-1 treated with OP50 or HB101. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 22–27. H The scatter plot shows the paralysis rate of Aβ or Aβ; pqm-1 treated with OP50 or HB101. One-way ANOVA followed by Tukey’s post hoc test, n = 4. I Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-4::GFP treated with OP50 or HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 19–22. J The scatter plot shows the paralysis rate of Aβ or Aβ; pek-1 treated with OP50 or HB101. One-way ANOVA followed by Tukey’s post hoc test, n = 3. K Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP treated with OP50 or HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 19–22. L The scatter plot showing the paralysis rate of Aβ or Aβ; sphk-1 treated with OP50 or HB101. One-way ANOVA followed by Tukey’s post hoc test, n = 3. M Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP and Phsp-6::GFP; atfs-1 treated with HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 28–47. N The scatter plot shows the paralysis rate of the Aβ AD model treated with the control HB101 + 0.02%DMSO or HB101 + MG-132. Two-tailed unpaired Student’s t-test. n = 6. O The scatter plot showing H2O2 resistance of HB101 in the Aβ AD model. Two-tailed unpaired Student’s t-test, n = 7. P The scatter plot showing H2O2 resistance of HB101 in the N2. Two-tailed unpaired Student’s t-test, n = 4. Q Representative fluorescence images and quantitative analysis of fluorescence intensity of Pgst-4::GFP, Two-tailed unpaired Student’s t-test, n = 24–28, Scale bar = 50 μm. R Representative fluorescence images and quantitative analysis of fluorescence intensity of Psod-3::GFP, Two-tailed unpaired Student’s t-test, n = 25–26, Scale bar = 50 μm. S Representative fluorescence images and quantitative analysis of fluorescence intensity of Pstr-1::GFP. The arrow in the diagram indicates neurons. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 25–29. T Representative fluorescence images and quantitative analysis of fluorescence intensity of Aβ; Peat-4::iGluSnFR. The arrow in the diagram indicates glutamate neurons. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 34–36. U Representative fluorescence images and quantitative analysis of fluorescence intensity of 488/405 nm excitation of Punc-30::NAD+ treated with OP50 or HB101. Two-tailed unpaired Student’s t-test. n = 20. Data are presented as mean ± SEM. ns, not significant (P > 0.05); ** P < 0.01; *** P < 0.001; **** P < 0.0001 HB101 mitigates Aβ aggregation by simultaneously enhancing lysosomal function, autophagy, UPRmt and UPRer. A Representative fluorescence images and quantitative analysis of the number of Aβ aggregation plaques treated with OP50 or HB101. Arrows indicate Aβ aggregation plaques. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 11–15. B Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP treated with HB101 or not. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 23–24. C Representative fluorescence images and quantitative analysis of fluorescence intensity of single lysosome treated with OP50 or HB101. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 51–135. D Representative fluorescence images and quantitative analysis of the number of autophagic substrates SQST-1::GFP treated with OP50 or HB101. Arrows indicate autophagic substrates. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 21–25. E Representative fluorescence images and quantitative analysis of the number of HLH-30::GFP positive nuclei treated with OP50 or HB101. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 30. F Representative fluorescence images and quantitative analysis of the total number of PHA-4::GFP positive nuclei at the pharyngeal bulb. Arrows indicate nucleus. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 21–23. G Representative fluorescence images and quantitative analysis of the red only puncta in the head of mCherry::GFP::LGG-1 treated with OP50 or HB101. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 22–27. H The scatter plot shows the paralysis rate of Aβ or Aβ; pqm-1 treated with OP50 or HB101. One-way ANOVA followed by Tukey’s post hoc test, n = 4. I Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-4::GFP treated with OP50 or HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 19–22. J The scatter plot shows the paralysis rate of Aβ or Aβ; pek-1 treated with OP50 or HB101. One-way ANOVA followed by Tukey’s post hoc test, n = 3. K Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP treated with OP50 or HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 19–22. L The scatter plot showing the paralysis rate of Aβ or Aβ; sphk-1 treated with OP50 or HB101. One-way ANOVA followed by Tukey’s post hoc test, n = 3. M Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP and Phsp-6::GFP; atfs-1 treated with HB101. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 28–47. N The scatter plot shows the paralysis rate of the Aβ AD model treated with the control HB101 + 0.02%DMSO or HB101 + MG-132. Two-tailed unpaired Student’s t-test. n = 6. O The scatter plot showing H2O2 resistance of HB101 in the Aβ AD model. Two-tailed unpaired Student’s t-test, n = 7. P The scatter plot showing H2O2 resistance of HB101 in the N2. Two-tailed unpaired Student’s t-test, n = 4. Q Representative fluorescence images and quantitative analysis of fluorescence intensity of Pgst-4::GFP, Two-tailed unpaired Student’s t-test, n = 24–28, Scale bar = 50 μm. R Representative fluorescence images and quantitative analysis of fluorescence intensity of Psod-3::GFP, Two-tailed unpaired Student’s t-test, n = 25–26, Scale bar = 50 μm. S Representative fluorescence images and quantitative analysis of fluorescence intensity of Pstr-1::GFP. The arrow in the diagram indicates neurons. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 25–29. T Representative fluorescence images and quantitative analysis of fluorescence intensity of Aβ; Peat-4::iGluSnFR. The arrow in the diagram indicates glutamate neurons. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 34–36. U Representative fluorescence images and quantitative analysis of fluorescence intensity of 488/405 nm excitation of Punc-30::NAD+ treated with OP50 or HB101. Two-tailed unpaired Student’s t-test. n = 20. Data are presented as mean ± SEM. ns, not significant (P > 0.05); ** P < 0.01; *** P < 0.001; **** P < 0.0001 As the primary cellular degradation center responsible for substrate recycling [20], we systematically assessed lysosomal function through multiple complementary approaches. The lysosomal surveillance response (LySR) pathway, characterized by enhanced lysosomal activity and protein aggregate clearance [21], was monitored using the LySR reporter strain Pcpr-5::GFP. HB101 treatment significantly increased Pcpr-5::GFP fluorescence intensity compared to OP50 controls (Fig. 2B), demonstrating LySR activation and suggesting HB101 promotes Aβ degradation through this pathway.‌ ‌We further evaluated lysosomal acidification using the pH-sensitive reporter NUC-1::pHTomato. HB101 treatment resulted in decreased NUC-1::pHTomato fluorescence (Fig. 2C), indicating enhanced lysosomal lumen acidification, a hallmark of increased lysosomal activity. Given autophagy’s critical role in degrading abnormal protein aggregates, we hypothesized that HB101 might enhance autophagic activity to reduce Aβ accumulation. However, quantification of SQST-1::GFP puncta, a well-established autophagic substrate and marker, revealed no significant reduction following HB101 treatment (Fig. 2D).‌ ‌To further investigate autophagic activity, we employed the MAH215 strain expressing mCherry::GFP::LGG-1 (C. elegans homolog of LC3). In this system, mature autolysosomes (ALs) appeared as mCherry-only positive puncta due to pH-dependent GFP quenching in acidic compartments [22]. Consistent with our expectations, HB101 treatment led to an increase in ALs number (Fig. 2G), providing evidence that it boosted autophagy activity. Notably, while HB101 did not alter the nuclear localization of HLH-30 (Fig. 2E), a master regulator of autophagy, it specifically promoted nuclear accumulation of PHA-4 (Fig. 2F), another key transcriptional regulator of autophagy. These findings collectively suggest that HB101 reduces Aβ aggregation through PHA-4-mediated autophagy activation. PQM-1, a zinc-finger transcription factor, plays a critical role in proteostasis in C. elegans. Mutation of pqm-1 abolished the protective effect of HB101 in the AD model (Fig. 2H), suggesting that PQM-1 is essential for mediating the protective function of HB101. Maintaining endoplasmic reticulum (ER) homeostasis is crucial for proper neuronal morphology [23] and its disruption activates the UPRer - an adaptive program that restores proteostasis by upregulating genes involved in protein folding and secretion [24]. To investigate whether UPRer contributes to Aβ reduction, we employed the Phsp-4::GFP reporter system to monitor UPRer activity. HB101 treatment significantly increased Phsp-4::GFP fluorescence intensity compared to OP50 controls (Fig. 2I), demonstrating UPRer activation.‌ ‌Given that PEK-1 is a master regulator of UPRer signaling, we tested its requirement using a pek-1 mutant strain. Complete abrogation of HB101’s protective effect in pek-1-deficient worms (Fig. 2J) confirmed that HB101 activates UPRer specifically through the PEK-1 pathway. The UPRmt is a stress response in mitochondria that triggers the transcriptional activation of mitochondrial chaperone proteins and proteases to preserve protein homeostasis [25]. To investigate whether the UPRmt contributes to Aβ reduction, we employed the Phsp-6::GFP reporter system to monitor UPRmt activity. HB101 treatment significantly increased Phsp-6::GFP fluorescence intensity compared to OP50 controls (Fig. 2K), demonstrating UPRmt activation. Given that ATFS-1 is a master regulator of UPRmt signaling [26], we crossed an atfs-1 mutant with Phsp-6::GFP to assess its requirement in HB101-mediated UPRmt activation. Notably, atfs-1 mutation abolished HB101-induced Phsp-6::GFP upregulation (Fig. 2M), indicating HB101’s dependence on this transcriptional regulator.‌ ‌Furthermore, since sphk-1 has been shown to mediate UPRmt signaling, we tested its involvement using a sphk-1 mutant. The impairment of HB101’s protective effect in sphk-1-deficient worms (Fig. 2L) confirmed that sphk-1 is required for HB101’s UPRmt-activating function. Since NAD+ has been shown to activate UPRmt and prevent age-associated decline [27], we investigated whether HB101 enhanced UPRmt by elevating NAD+ levels. To monitor NAD+ dynamics in vivo, we generated a transgenic strain expressing an NAD+ biosensor [28], where increased fluorescence indicates a lower NAD+ level. We implemented normalization with 405 nm excitation and quantified the 488/405 nm ratio, as its 405 nm excitation peak is insensitive to NAD+. However, HB101 treatment did not alter the D-type motor neuron-specific biosensor’s fluorescence intensity (Fig. 2U), demonstrating no significant change in NAD+ levels.‌ These results collectively indicate that HB101 activates UPRmt through an NAD+-independent mechanism. Given the essential role of the proteasome in protein degradation, we hypothesized that HB101 might enhance proteasome activity to facilitate the clearance of Aβ aggregates. To test this hypothesis, we treated worms with HB101 in combination with the proteasome inhibitor MG-132. Notably, pharmacological inhibition of the proteasome did not significantly abrogate the protective effects of HB101 (Fig. 2N), indicating that HB101 exerts its protective function independently of proteasome-mediated degradation pathways. Under H2O2-induced oxidative stress, HB101 conferred no significant protective effect on wild-type N2 worms compared with OP50, but exerted a marked protective effect on Aβ-expressing AD model worms (Fig. 2O-P). Glutathione S-transferase (GST-4)-mediated pathways represent promising therapeutic targets for neurodegenerative diseases [29]. HB101 treatment significantly increased Pgst-4::GFP fluorescence intensity compared to OP50 controls (Fig. 2Q), demonstrating enhanced GST-4 expression may contribute to Aβ toxicity reduction. Consistently, HB101 also upregulated Psod-3::GFP expression (Fig. 2R), encoding a superoxide dismutase enzyme. These results suggest that HB101 may enhance oxidative stress resistance via GST-4 and SOD-3. While Bacillus subtilis has been shown to produce neuroprotective nitric oxide (NO) against α-syn aggregation [30], our NO sensor analysis using Pstr-1::GFP revealed no significant difference between HB101 and OP50 treatments (Fig. 2S), indicating NO-independent mechanisms. Additionally, we investigated potential glutamate modulation using an iGluSnFR sensor strain [31], finding no HB101-induced changes in iGluSnFR intensity (Fig. 2T), thereby excluding glutamate concentration regulation as a mechanism of action. Our data suggest that HB101 exerts its functions through multiple signaling pathways. To further explore the regulatory relationships among these pathways, we investigated their potential crosstalk. First, we treated Phsp-6::GFP reporter worms with the autophagy activator rapamycin, and the results showed that rapamycin had no obvious effect on Phsp-6::GFP expression (Fig. 3A). In addition, rapamycin treatment reduced the expression level of Pcpr-5::GFP (Fig. 3B). We next used the atfs-1 mutant to inhibit the UPRmt and examined its influence on autophagy. We found that the atfs-1 mutation did not block the HB101-mediated enhancement of autophagy (Fig. 3C). However, loss of atfs-1 function impaired the regulatory effect of HB101 on Pcpr-5::GFP (Fig. 3D). Fig. 3Pathway analysis. A Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP treated with OP50 or Rapamycin. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 35. B Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP treated with OP50 or Rapamycin. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 35. C Representative fluorescence images and quantitative analysis of the red only puncta in the head of mCherry::GFP::LGG-1 and mCherry::GFP::LGG-1; atfs-1 treated with OP50 or HB101. Scale bar = 50 μm. One-way ANOVA followed by Tukey’s post hoc test, n = 20. D Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP and Pcpr-5::GFP; atfs-1 treated with OP50 or HB101. Scale bar = 50 μm. One-way ANOVA followed by Tukey’s post hoc test, n = 29–30. Data are presented as mean ± SEM. ns, not significant (P > 0.05); *** P < 0.001; **** P < 0.0001 Pathway analysis. A Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP treated with OP50 or Rapamycin. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 35. B Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP treated with OP50 or Rapamycin. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test, n = 35. C Representative fluorescence images and quantitative analysis of the red only puncta in the head of mCherry::GFP::LGG-1 and mCherry::GFP::LGG-1; atfs-1 treated with OP50 or HB101. Scale bar = 50 μm. One-way ANOVA followed by Tukey’s post hoc test, n = 20. D Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP and Pcpr-5::GFP; atfs-1 treated with OP50 or HB101. Scale bar = 50 μm. One-way ANOVA followed by Tukey’s post hoc test, n = 29–30. Data are presented as mean ± SEM. ns, not significant (P > 0.05); *** P < 0.001; **** P < 0.0001 Untargeted metabolomic profiling of HB101 To identify differential metabolites between HB101 and the control OP50, we conducted untargeted metabolomic analysis. Principal component analysis (PCA) demonstrated significant inter-group variability, clearly separating the OP50 and HB101 cohorts (Fig. 4A), suggesting substantial differences in their global metabolic profiles. Orthogonal partial least squares discriminant analysis (OPLS-DA) further validated the distinct separation between these groups (Fig. 4B). The volcano plot analysis revealed 130 significantly upregulated and 89 downregulated metabolites (Fig. 4C-E). Moreover, hierarchical clustering of differential metabolites exhibited distinct metabolic patterns, with clear segregation between the HB101 (HB101-1-HB101-3) and OP50 (OP50-1-OP50-3) groups (Fig. 4F), further confirming the pronounced metabolic divergence between the two bacterial strains. Fig. 4Untargeted metabolomic profiling of HB101. A Score plot of principal component analysis (PCA). B Orthogonal partial least squares-discriminant analysis (OPLS-DA) plot. C Volcano plot comparing the OP50 and HB101 groups, with red representing upregulated metabolites and green representing downregulated metabolites. D Violin plot of the relative content of glutaric acid (glutarate) in OP50 and HB101 groups. E Bar chart displaying the fold changes of the top 21 metabolites ranked by their fold change values. F Heatmap of differential metabolite expression between the HB101 group (HB101-1-HB101-3) and the OP50 group (OP50-1-OP50-3). ** P < 0.01 Untargeted metabolomic profiling of HB101. A Score plot of principal component analysis (PCA). B Orthogonal partial least squares-discriminant analysis (OPLS-DA) plot. C Volcano plot comparing the OP50 and HB101 groups, with red representing upregulated metabolites and green representing downregulated metabolites. D Violin plot of the relative content of glutaric acid (glutarate) in OP50 and HB101 groups. E Bar chart displaying the fold changes of the top 21 metabolites ranked by their fold change values. F Heatmap of differential metabolite expression between the HB101 group (HB101-1-HB101-3) and the OP50 group (OP50-1-OP50-3). ** P < 0.01 The metabolite glutarate, derived from HB101, mitigates AD model phenotypes by simultaneously enhancing lysosomal function, autophagy and UPRmt To identify the functional metabolites of HB101, we screened for metabolites exhibiting significantly higher expression in HB101 compared to the control strain OP50. Among the over 30 candidates analyzed (Fig. 5A, not list all of them), we identified glutarate as a candidate metabolite. At a concentration of 10 µM, glutarate significantly reduced paralysis in the Aβ AD model, while 50 µM also demonstrated a significant reduction effect (Fig. 5B). Metabolomic quantification revealed a substantial 11.41-fold increase in glutarate levels in the HB101 group compared to OP50 (Fig. 4D, E). Notably, 10 µM glutarate administration improved learning and memory deficits in the GRU102 AD model, as evidenced by a reduced number of touches (Fig. 5C). Furthermore, 10 µM glutarate restored commissure numbers to near-physiological levels in the AD model (Fig. 5D), providing robust evidence for its neuroprotective efficacy against Aβ-induced neurodegeneration. Fig. 5The metabolite glutarate mitigates AD model phenotypes by simultaneously enhancing lysosomal function, autophagy, and UPRmt. A Bar graph showing the paralysis rate of the Aβ AD model treated with small-molecule metabolites of HB101. Fisher's exact test, n = 25–30. B The scatter plot showing the paralysis rate of the Aβ AD model treated with different concentrations of glutarate or not. One-way ANOVA followed by Tukey’s post hoc test. n = 4. C Scatter plot showing the number of retreats treated with glutarate or not. One-way ANOVA followed by Tukey’s post hoc test. n = 17–21. D Representative fluorescence images and quantitative analysis of commissures in the Aβ AD model and the control. Arrows indicate commissures. Scale bar = 50 μm. One-way ANOVA followed by Tukey’s post hoc test. n = 18–24. E Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP treated with glutarate or not. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 21–23. F Representative fluorescence images and quantitative analysis of fluorescence intensity of single lysosome treated with OP50 or glutarate. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 110–127. G-H Representative fluorescence images and quantitative analysis of the red only puncta in the head of mCherry::GFP::LGG-1 treated with OP50 or glutarate. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 24–29. I Representative fluorescence images and quantitative analysis of the total number of PHA-4::GFP positive nuclei at the pharyngeal bulb. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 25. J Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP treated with glutarate or not. Two-tailed unpaired Student’s t-test. n = 15–21. K Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP and Phsp-6::GFP; atfs-1 treated with glutarate. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 30–32. L Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-4::GFP treated with glutarate or not. Two-tailed unpaired Student’s t-test. bar = 50 μm. n = 15–21. Data are presented as mean ± SEM. ns, not significant (P > 0.05); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 The metabolite glutarate mitigates AD model phenotypes by simultaneously enhancing lysosomal function, autophagy, and UPRmt. A Bar graph showing the paralysis rate of the Aβ AD model treated with small-molecule metabolites of HB101. Fisher's exact test, n = 25–30. B The scatter plot showing the paralysis rate of the Aβ AD model treated with different concentrations of glutarate or not. One-way ANOVA followed by Tukey’s post hoc test. n = 4. C Scatter plot showing the number of retreats treated with glutarate or not. One-way ANOVA followed by Tukey’s post hoc test. n = 17–21. D Representative fluorescence images and quantitative analysis of commissures in the Aβ AD model and the control. Arrows indicate commissures. Scale bar = 50 μm. One-way ANOVA followed by Tukey’s post hoc test. n = 18–24. E Representative fluorescence images and quantitative analysis of fluorescence intensity of Pcpr-5::GFP treated with glutarate or not. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 21–23. F Representative fluorescence images and quantitative analysis of fluorescence intensity of single lysosome treated with OP50 or glutarate. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 110–127. G-H Representative fluorescence images and quantitative analysis of the red only puncta in the head of mCherry::GFP::LGG-1 treated with OP50 or glutarate. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 24–29. I Representative fluorescence images and quantitative analysis of the total number of PHA-4::GFP positive nuclei at the pharyngeal bulb. Scale bar = 20 μm. Two-tailed unpaired Student’s t-test. n = 25. J Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP treated with glutarate or not. Two-tailed unpaired Student’s t-test. n = 15–21. K Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-6::GFP and Phsp-6::GFP; atfs-1 treated with glutarate. Scale bar = 50 μm. Two-tailed unpaired Student’s t-test. n = 30–32. L Representative fluorescence images and quantitative analysis of fluorescence intensity of Phsp-4::GFP treated with glutarate or not. Two-tailed unpaired Student’s t-test. bar = 50 μm. n = 15–21. Data are presented as mean ± SEM. ns, not significant (P > 0.05); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 Building on our finding that HB101 may function by enhancing lysosomal activity, we hypothesized that glutarate operates through a similar mechanism. Intriguingly, glutarate treatment significantly increased Pcpr-5::GFP fluorescence intensity compared to OP50 controls (Fig. 5E), demonstrating activation of the LySR. Furthermore, glutarate treatment led to decreased NUC-1::pHTomato fluorescence (Fig. 5F), indicating enhanced lysosomal lumen acidification, a hallmark of increased lysosomal activity. Since HB101 enhanced autophagy activity, we hypothesized that glutarate functions similarly. As expected, glutarate treatment significantly increased the number of ALs (Fig. 5G, H), confirming its autophagy-enhancing effect. Intriguingly, glutarate promoted nuclear accumulation of PHA-4 (Fig. 5I), a key transcriptional regulator of autophagy. These findings collectively suggest that glutarate ameliorates AD-associated phenotypes through PHA-4-mediated autophagy activation. We have demonstrated that HB101 may act by activating the UPRmt, and we assessed the Phsp-6::GFP reporter, which serves as a UPRmt indicator, following glutarate treatment. The results indicated that glutarate enhanced Phsp-6::GFP expression (Fig. 5J), implying that glutarate operates through UPRmt. Notably, atfs-1 mutation impaired glutarate-induced Phsp-6::GFP upregulation (Fig. 5K), indicating glutarate’s dependence on this transcriptional regulator. HB101 may function via activation of the UPRer, so we evaluated the Phsp-4::GFP reporter, a UPRer sensor, after glutarate treatment. The data revealed that glutarate could not increase Phsp-4::GFP expression (Fig. 5L), suggesting that glutarate does not act through UPRer. Discussion Probiotics have emerged as a promising therapeutic strategy for AD, yet the limited number of investigated strains and the incomplete understanding of their mechanisms remain significant challenges. In this study, we demonstrated that E. coli HB101 ameliorates AD-like phenotypes in a model system and systematically elucidated its therapeutic mechanisms. Our findings reveal that the neuroprotective effects of HB101 are contingent upon viable bacterial cells, as heat-killed preparations (autoclaved or pasteurized) completely lost efficacy. This observation suggests two plausible explanations: (1) the protective mechanism requires live bacterial metabolic activity, or (2) bioactive components are thermolabile and irreversibly denatured by heat treatment. Further studies are warranted to distinguish between these possibilities. HB101 mitigates Aβ aggregation through a multifaceted mechanism. First, it activates the LySR, a known mediator of protein aggregate clearance. Second, it enhances lysosomal acidification, indicative of elevated lysosomal activity, potentially via modulation of the lysosomal membrane proton pump, a hypothesis that merits further investigation. Third, HB101 activates the UPRmt and UPRer, two critical pathways for maintaining proteostasis. Collectively, these findings suggest that HB101 exerts its anti-aggregation effects through a synergistic interplay of pathways, likely mediated by diverse functional metabolites targeting distinct cellular processes. It is well recognized that pharmacological agents usually exert their functions through multiple parallel pathways, and extensive crosstalk exists among these pathways, which is biologically reasonable. We found that HB101 functions via several distinct regulatory pathways. To clarify the relationship among these pathways, we selectively interfered with one pathway and subsequently detected the readouts of the others. Our results indicate that most of these pathways operate largely independently and may function in parallel. Intriguingly, we found that autophagy activation affects the lysosomal surveillance pathway, and inhibition of the UPRmt also influences lysosomal surveillance, suggesting evident crosstalk between these pathways. The underlying mechanism remains to be further investigated in follow-up studies. The therapeutic functions of probiotics are typically attributed to their metabolites. However, the vast chemical diversity of metabolites has posed a major challenge in identifying key active components. Here, we employed an efficient C. elegans screening model coupled with metabolomic analysis to identify glutarate as a potential metabolite of HB101. Mechanistic studies demonstrated that glutarate mimics key effects of HB101, including the activation of the LySR, UPRmt, enhancement of lysosomal acidification. However, a subtle distinction emerged: glutarate does not activate the UPRer pathways, which are part of the HB101’s mechanism. This divergence suggests that HB101’s full therapeutic potential may be mediated by additional, yet-to-be-identified metabolites. Future research should systematically screen a broader panel of metabolites to fully elucidate the complete mechanistic landscape. Our study identified a novel bacterial metabolite that effectively ameliorates the pathological phenotypes in AD models. This finding broadens the repertoire of neuroprotective bacterial metabolites, as previously reported protective metabolites are predominantly confined to short-chain fatty acids, indole derivatives, and bile acids. Moreover, our work further expands the mechanistic landscape underlying the anti-AD effects of microbial metabolites. For the first time, we report that E. coli HB101 alleviates AD-like symptoms in a model system. Our mechanistic investigations highlight its role in enhancing lysosomal activity and activating UPRmt/UPRer pathways. By leveraging a rapid screening approach, we successfully identified glutarate might be one of the metabolites of HB101, which exerts effects through mechanisms partially overlapping with but distinct from those of the parent strain. This study not only advances our understanding of probiotic mechanisms but also provides a valuable framework for identifying active metabolites from other probiotics. The insights gained here may inform the development of novel AD therapeutics, particularly those targeting protein homeostasis and lysosomal function.